Server Status window work

Hi,

The roadmap still shows I need to work on the server status window. But
it doesn't explain what I have in mind. Dave gave me some hints and
todos. Here is my todo list on this :

* Replace the wxNotebook widget with the wxAuiNotebook so users can
close tabs and rearrange tabs order via drag-and-drop.

* Add the possibility to open many Server Status windows (we can
already do this with the SQL query tool) so a user can see many
reports at the same time.

* Add a dockable toolbar to enable changing connection server, timer
slider, filter and sort options.

* Add graph reports for different statistics views, something like
the Gnome System Monitor but specific to PostgreSQL (transaction
rate, read/hit ratio, number of connections, bgwriter stats come to
mind).

* Better display of the logfile (probably using the log_line_prefix
option to know which informations are available).

* Command line options to start the Server Status window on it's own,
much as we can already do with the query tool.

* Colour coding for the backend lines. Say green for running OK, orange
for running, but with the query time being past a configurable limit,
blue for idle, and red when waiting for a lock etc.

* The ability to select a running backend, and copy the query straight
into a new query tool window (caveat: the stats command string may be
truncated, and still contain parameter placeholders).

* Ability to select a backend process and display just the locks
it's holding, and the one it's waiting on.

* (bit more complex) Display the locks in a logical tree rather than a
list (applies to the previous point as well as the current tab).

The last five are Dave's ideas. I just copy/paste one of his emails.

I know this is a huge todo list and I don't pretend doing everything
alone and for the next release. Actually, I think I'll do one patch for
each todo. I hope to begin the work on this next week.

Comments? Ideas?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Sun, Jun 15, 2008 at 8:45 AM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Hi,
>
> The roadmap still shows I need to work on the server status window. But it
> doesn't explain what I have in mind. Dave gave me some hints and todos. Here
> is my todo list on this :
>

Snip todo list...

> The last five are Dave's ideas. I just copy/paste one of his emails.
>
> I know this is a huge todo list and I don't pretend doing everything alone
> and for the next release. Actually, I think I'll do one patch for each todo.
> I hope to begin the work on this next week.
>
> Comments? Ideas?

I'm glad you're still planning to work on this :-). I'd be happy to
see this being achieved over time in multiple patches.

I won't comment on the actual todo's - we already discussed them as you know.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Hi,

Guillaume Lelarge a écrit :
> [...]
> The roadmap still shows I need to work on the server status window. But
> it doesn't explain what I have in mind. Dave gave me some hints and
> todos.

At last, I'm working on this.

> Here is my todo list on this :
> [...]
> * Command line options to start the Server Status window on it's own,
> much as we can already do with the query tool.
>

The patch attached implements this. It seems I just copy and paste the
query tool code, and in fact it's true. Not sure if I should refactor
that code. It'll be less readable.

Regards.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Sun, Jan 4, 2009 at 12:53 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Hi,
>
> Guillaume Lelarge a écrit :
>> [...]
>> The roadmap still shows I need to work on the server status window. But
>> it doesn't explain what I have in mind. Dave gave me some hints and
>> todos.
>
> At last, I'm working on this.

You don't have long - PG beta is planned for February.

>> Here is my todo list on this :
>> [...]
>> * Command line options to start the Server Status window on it's own,
>> much as we can already do with the query tool.
>>
>
> The patch attached implements this. It seems I just copy and paste the
> query tool code, and in fact it's true. Not sure if I should refactor
> that code. It'll be less readable.

Yeah - I'd be inclined to leave it. It's not clear how it could be
usefully re-jiggered.

Please commit if you're happy it tests OK. It looks reasonable to my eyes.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Sun, Jan 4, 2009 at 12:53 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> Hi,
>>
>> Guillaume Lelarge a écrit :
>>> [...]
>>> The roadmap still shows I need to work on the server status window. But
>>> it doesn't explain what I have in mind. Dave gave me some hints and
>>> todos.
>> At last, I'm working on this.
>
> You don't have long - PG beta is planned for February.
>

Which means pgAdmin too ? or do we have some more time ? this will be
pretty hard to add support for PostgreSQL new features in such a small
delay.

>>> Here is my todo list on this :
>>> [...]
>>> * Command line options to start the Server Status window on it's own,
>>> much as we can already do with the query tool.
>>>
>> The patch attached implements this. It seems I just copy and paste the
>> query tool code, and in fact it's true. Not sure if I should refactor
>> that code. It'll be less readable.
>
> Yeah - I'd be inclined to leave it. It's not clear how it could be
> usefully re-jiggered.
>
> Please commit if you're happy it tests OK. It looks reasonable to my eyes.
>

Done, thanks.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Mon, Jan 5, 2009 at 10:48 AM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Which means pgAdmin too ? or do we have some more time ? this will be
> pretty hard to add support for PostgreSQL new features in such a small
> delay.

We need to go to beta at the same time or before, because numerous
installers include both pgAdmin and PostgreSQL. The support issue is a
valid problem, which I bleated about on -hackers yesterday :-). Ashesh
is working on such issues, and I expect to put in some time myself.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Mon, Jan 5, 2009 at 10:48 AM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> Which means pgAdmin too ? or do we have some more time ? this will be
>> pretty hard to add support for PostgreSQL new features in such a small
>> delay.
>
> We need to go to beta at the same time or before, because numerous
> installers include both pgAdmin and PostgreSQL.

Ouch, I always forgot about this. Sorry.

> The support issue is a
> valid problem, which I bleated about on -hackers yesterday :-).

Yeah, I've read this yesterday.

> Ashesh is working on such issues, and I expect to put in some time myself.
>

Great. I suppose january will be pgAdmin's month for me too. Server
status is on top. And february will be translation's month (pgAdmin .po,
PostgreSQL .po, and PostgreSQL manual).

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Mon, Jan 5, 2009 at 11:16 AM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> Great. I suppose january will be pgAdmin's month for me too. Server
> status is on top. And february will be translation's month (pgAdmin .po,
> PostgreSQL .po, and PostgreSQL manual).

Yup, sounds about right :-). Any joy with that dialogue issue yet?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Mon, Jan 5, 2009 at 11:16 AM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> Great. I suppose january will be pgAdmin's month for me too. Server
>> status is on top. And february will be translation's month (pgAdmin .po,
>> PostgreSQL .po, and PostgreSQL manual).
>
> Yup, sounds about right :-). Any joy with that dialogue issue yet?
>

No. I worked a lot on it and couldn't find a way to make it better. I
removed the panels and it kind of works on Linux, but not at all on Mac.
I'm a bit worried about this one.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Mon, Jan 5, 2009 at 12:31 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Dave Page a écrit :
>> On Mon, Jan 5, 2009 at 11:16 AM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>>
>>> Great. I suppose january will be pgAdmin's month for me too. Server
>>> status is on top. And february will be translation's month (pgAdmin .po,
>>> PostgreSQL .po, and PostgreSQL manual).
>>
>> Yup, sounds about right :-). Any joy with that dialogue issue yet?
>>
>
> No. I worked a lot on it and couldn't find a way to make it better. I
> removed the panels and it kind of works on Linux, but not at all on Mac.
> I'm a bit worried about this one.

Can you post what you have please?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Mon, Jan 5, 2009 at 12:31 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> Dave Page a écrit :
>>> On Mon, Jan 5, 2009 at 11:16 AM, Guillaume Lelarge
>>> <guillaume(at)lelarge(dot)info> wrote:
>>>
>>>> Great. I suppose january will be pgAdmin's month for me too. Server
>>>> status is on top. And february will be translation's month (pgAdmin .po,
>>>> PostgreSQL .po, and PostgreSQL manual).
>>> Yup, sounds about right :-). Any joy with that dialogue issue yet?
>>>
>> No. I worked a lot on it and couldn't find a way to make it better. I
>> removed the panels and it kind of works on Linux, but not at all on Mac.
>> I'm a bit worried about this one.
>
> Can you post what you have please?
>

Sure, no problem. The last patch I was working on is the one attached,
though it seems to work better on Mac than on the linux box... grmbl...

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Hi,

Guillaume Lelarge a écrit :
> [...]
> The roadmap still shows I need to work on the server status window. But
> it doesn't explain what I have in mind. Dave gave me some hints and
> todos. Here is my todo list on this :
>
> * Replace the wxNotebook widget with the wxAuiNotebook so users can
> close tabs and rearrange tabs order via drag-and-drop.
>

This is done in the attached patch. frmStatus.xrc should be removed, all
the widgets are created in frmStatus.cpp.

Please, do not commit, it needs more testing and, I think, more work.
And debug too: in standalone mode, it crashes at exit.

> * Add the possibility to open many Server Status windows (we can
> already do this with the SQL query tool) so a user can see many
> reports at the same time.
>

Works too.

> * Add a dockable toolbar to enable changing connection server, timer
> slider, filter and sort options.
>

I'm wondering if I should add this and a menubar. My idea is to work
with panes and make it possible to add many panes in a same window. Each
pane will handle one type of report : activity, locks, transactions,
logfile.

> [...]
> * Colour coding for the backend lines. Say green for running OK, orange
> for running, but with the query time being past a configurable limit,
> blue for idle, and red when waiting for a lock etc.
>

Dave, this one is from you. Where do you want to put the configuration
of the server status. In the options dialog already available? I add a
notebookpage with all server status parameters?

> * The ability to select a running backend, and copy the query straight
> into a new query tool window (caveat: the stats command string may be
> truncated, and still contain parameter placeholders).
>
> * Ability to select a backend process and display just the locks
> it's holding, and the one it's waiting on.
>

For these two, a contextual menu seems a good choice. Other ideas?

If you have comments, ideas, requests, please speak aloud :)

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Hi,
>
> Guillaume Lelarge a écrit :
>> [...]
>> The roadmap still shows I need to work on the server status window. But
>> it doesn't explain what I have in mind. Dave gave me some hints and
>> todos. Here is my todo list on this :
>>
>> * Replace the wxNotebook widget with the wxAuiNotebook so users can
>> close tabs and rearrange tabs order via drag-and-drop.
>>
>
> This is done in the attached patch. frmStatus.xrc should be removed, all
> the widgets are created in frmStatus.cpp.
>
> Please, do not commit, it needs more testing and, I think, more work.
> And debug too: in standalone mode, it crashes at exit.
>

Sorry this is not the last patch. It misses a #include of wx/aui/auibook.h.

I will work on it tonight but I would appreciate any
ideas/comments/requests.

--
Guillaume.

On Mon, Jan 5, 2009 at 11:26 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

>> * Replace the wxNotebook widget with the wxAuiNotebook so users can
>> close tabs and rearrange tabs order via drag-and-drop.
>>
>
> This is done in the attached patch. frmStatus.xrc should be removed, all
> the widgets are created in frmStatus.cpp.

Doesn't seem to make much difference on OS X. I certainly can't
rearrange tabs etc.

>> * Add the possibility to open many Server Status windows (we can
>> already do this with the SQL query tool) so a user can see many
>> reports at the same time.
>>
>
> Works too.

Yup.

>> * Add a dockable toolbar to enable changing connection server, timer
>> slider, filter and sort options.
>>
>
> I'm wondering if I should add this and a menubar. My idea is to work
> with panes and make it possible to add many panes in a same window. Each
> pane will handle one type of report : activity, locks, transactions,
> logfile.

Yep, could do - even if it just has a view & file menu.

>> * Colour coding for the backend lines. Say green for running OK, orange
>> for running, but with the query time being past a configurable limit,
>> blue for idle, and red when waiting for a lock etc.
>>
>
> Dave, this one is from you. Where do you want to put the configuration
> of the server status. In the options dialog already available? I add a
> notebookpage with all server status parameters?

That would seem sensible

>> * The ability to select a running backend, and copy the query straight
>> into a new query tool window (caveat: the stats command string may be
>> truncated, and still contain parameter placeholders).
>>
>> * Ability to select a backend process and display just the locks
>> it's holding, and the one it's waiting on.
>>
>
> For these two, a contextual menu seems a good choice. Other ideas?

Potentially difficult to use without a mouse - but you could add a
top-level menu option as well.

/D

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Mon, Jan 5, 2009 at 11:26 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>>> * Replace the wxNotebook widget with the wxAuiNotebook so users can
>>> close tabs and rearrange tabs order via drag-and-drop.
>>>
>> This is done in the attached patch. frmStatus.xrc should be removed, all
>> the widgets are created in frmStatus.cpp.
>
> Doesn't seem to make much difference on OS X. I certainly can't
> rearrange tabs etc.
>

Works for me on Linux and Mac OS X. I can close tabs and rearrange them.

The new patch attached fixes the crash issue I had yesterday. It also
adds a close box on each tab and customizes the wxAuiNotebook style.

>>> * Add the possibility to open many Server Status windows (we can
>>> already do this with the SQL query tool) so a user can see many
>>> reports at the same time.
>>>
>> Works too.
>
> Yup.
>
>>> * Add a dockable toolbar to enable changing connection server, timer
>>> slider, filter and sort options.
>>>
>> I'm wondering if I should add this and a menubar. My idea is to work
>> with panes and make it possible to add many panes in a same window. Each
>> pane will handle one type of report : activity, locks, transactions,
>> logfile.
>
> Yep, could do - even if it just has a view & file menu.
>

OK.

>>> * Colour coding for the backend lines. Say green for running OK, orange
>>> for running, but with the query time being past a configurable limit,
>>> blue for idle, and red when waiting for a lock etc.
>>>
>> Dave, this one is from you. Where do you want to put the configuration
>> of the server status. In the options dialog already available? I add a
>> notebookpage with all server status parameters?
>
> That would seem sensible
>

OK.

>>> * The ability to select a running backend, and copy the query straight
>>> into a new query tool window (caveat: the stats command string may be
>>> truncated, and still contain parameter placeholders).
>>>
>>> * Ability to select a backend process and display just the locks
>>> it's holding, and the one it's waiting on.
>>>
>> For these two, a contextual menu seems a good choice. Other ideas?
>
> Potentially difficult to use without a mouse - but you could add a
> top-level menu option as well.
>

I can also add shortcut keys.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> On Mon, Jan 5, 2009 at 11:26 PM, Guillaume Lelarge
>>[...]
>>>> * Add a dockable toolbar to enable changing connection server, timer
>>>> slider, filter and sort options.
>>>>
>>> I'm wondering if I should add this and a menubar. My idea is to work
>>> with panes and make it possible to add many panes in a same window. Each
>>> pane will handle one type of report : activity, locks, transactions,
>>> logfile.
>> Yep, could do - even if it just has a view & file menu.
>>
>
> OK.
>

Thinking again about it, a view menu would be the only way to re-add a
tab that has been closed previously. (in the current patch, the only way
to re-open a tab previously closed is to close and open the server
status dialog)

One last question: should we record the new display (ie tabs closed and
rearranged) so we can completely restore it next time the user opens the
server status dialog?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Tue, Jan 6, 2009 at 10:30 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Dave Page a écrit :
>> On Mon, Jan 5, 2009 at 11:26 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>>
>> Doesn't seem to make much difference on OS X. I certainly can't
>> rearrange tabs etc.
>>
>
> Works for me on Linux and Mac OS X. I can close tabs and rearrange them.

Oh, I see - you've got all the cool functionality turned off :-(

> The new patch attached fixes the crash issue I had yesterday.

Not entirely - if you close the Log tab, it seems to crash when
pgAdmin is closed.

> It also
> adds a close box on each tab and customizes the wxAuiNotebook style.

Let's get it to work with wxAUI_NB_DEFAULT_STYLE. That allows the
split functionality which is what I think makes this change worthwhile
- the ability to view more than one tab at once and arrange them as
required. A quick test shows it needs a new way to figure out which
tabs are visible (and therefore need refreshing), and there are some
sizing issues which appear to all be our old friend the generic
wxListCtrl.

It also looks a bit nicer than having the close box on all tabs I think.

> Thinking again about it, a view menu would be the only way to re-add a
> tab that has been closed previously. (in the current patch, the only way
> to re-open a tab previously closed is to close and open the server
> status dialog)

Yeah - plus it's where we put the option to reset the layout to the
default view.

> One last question: should we record the new display (ie tabs closed and
> rearranged) so we can completely restore it next time the user opens the
> server status dialog?

Yes - but isn't that handled by the AUI code and associated functions
as per frmMain etc?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Tue, Jan 6, 2009 at 10:30 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> Dave Page a écrit :
>>> On Mon, Jan 5, 2009 at 11:26 PM, Guillaume Lelarge
>>> <guillaume(at)lelarge(dot)info> wrote:
>>>
>>> Doesn't seem to make much difference on OS X. I certainly can't
>>> rearrange tabs etc.
>>>
>> Works for me on Linux and Mac OS X. I can close tabs and rearrange them.
>
> Oh, I see - you've got all the cool functionality turned off :-(
>

I suppose the cool functionality is wxAUI_NB_TAB_SPLIT :)

>> The new patch attached fixes the crash issue I had yesterday.
>
> Not entirely - if you close the Log tab, it seems to crash when
> pgAdmin is closed.
>

You're right. The new version fixes this.

>> It also
>> adds a close box on each tab and customizes the wxAuiNotebook style.
>
> Let's get it to work with wxAUI_NB_DEFAULT_STYLE. That allows the
> split functionality which is what I think makes this change worthwhile
> - the ability to view more than one tab at once and arrange them as
> required. A quick test shows it needs a new way to figure out which
> tabs are visible (and therefore need refreshing), and there are some
> sizing issues which appear to all be our old friend the generic
> wxListCtrl.
>

Actually, I find this great. It works nice on Mac OS X, but I disabled
it on v2 because it crashes the linux build. Perhaps I did something
wrong but even the auidemo sample from wxGTK crashes.

(re-enables the "cool functionality", so crashes a lot on my laptop)

> It also looks a bit nicer than having the close box on all tabs I think.
>

You mean, to have a close box only on the active tab? (aka
wxAUI_NB_CLOSE_ON_ACTIVE_TAB)

>> Thinking again about it, a view menu would be the only way to re-add a
>> tab that has been closed previously. (in the current patch, the only way
>> to re-open a tab previously closed is to close and open the server
>> status dialog)
>
> Yeah - plus it's where we put the option to reset the layout to the
> default view.
>

OK.

>> One last question: should we record the new display (ie tabs closed and
>> rearranged) so we can completely restore it next time the user opens the
>> server status dialog?
>
> Yes - but isn't that handled by the AUI code and associated functions
> as per frmMain etc?
>

With the perspective code, yes. I need to better understand it to code
it for the server status window.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Wed, Jan 7, 2009 at 10:08 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> I suppose the cool functionality is wxAUI_NB_TAB_SPLIT :)

Yep.

> Actually, I find this great. It works nice on Mac OS X, but I disabled
> it on v2 because it crashes the linux build. Perhaps I did something
> wrong but even the auidemo sample from wxGTK crashes.
>
> (re-enables the "cool functionality", so crashes a lot on my laptop)

Oh, rats. I hate getting my hands dirty on wxGTK :-(

> You mean, to have a close box only on the active tab? (aka
> wxAUI_NB_CLOSE_ON_ACTIVE_TAB)

Yes.

> With the perspective code, yes. I need to better understand it to code
> it for the server status window.

It's pretty simple really - you've got most of it, just the toggle
functions to add to switch things on or off via the menu. We also used
to add event handlers for OnSize and OnEraseBackground to fix some
display issues, but I'm not sure if thats still a problem with moden
versions of wxWidgets.

Some comments on the patch:

- The form setup code is a pretty large hunk of code. Can we get some
more comments and blank lines in there to break things up a little?
- The form underneath the AUI controls seems to have some drawing
issues on XP. The refresh bar is just about unusable for example.

Thats about it for the moment - nothing too serious.

BTW, any idea why we use a listctrl to display the log? Any thoughts
on changing it to a text area?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Wed, Jan 7, 2009 at 10:08 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> I suppose the cool functionality is wxAUI_NB_TAB_SPLIT :)
>
> Yep.
>
>> Actually, I find this great. It works nice on Mac OS X, but I disabled
>> it on v2 because it crashes the linux build. Perhaps I did something
>> wrong but even the auidemo sample from wxGTK crashes.
>>
>> (re-enables the "cool functionality", so crashes a lot on my laptop)
>
> Oh, rats. I hate getting my hands dirty on wxGTK :-(
>

I found this bug report : http://trac.wxwidgets.org/ticket/4841

My issue seems related to this.

>> You mean, to have a close box only on the active tab? (aka
>> wxAUI_NB_CLOSE_ON_ACTIVE_TAB)
>
> Yes.
>
>> With the perspective code, yes. I need to better understand it to code
>> it for the server status window.
>
> It's pretty simple really - you've got most of it, just the toggle
> functions to add to switch things on or off via the menu. We also used
> to add event handlers for OnSize and OnEraseBackground to fix some
> display issues, but I'm not sure if thats still a problem with moden
> versions of wxWidgets.
>

Done.

> Some comments on the patch:
>
> - The form setup code is a pretty large hunk of code. Can we get some
> more comments and blank lines in there to break things up a little?

It should be done. There's more comments and the code is a bit factored.

> - The form underneath the AUI controls seems to have some drawing
> issues on XP. The refresh bar is just about unusable for example.
>

Yeah. Even with this new patch, I still have this issue and no idea how
to fix this. I tried the OnSize and OnEraseBackground methods, but that
didn't fix it.

> Thats about it for the moment - nothing too serious.
>

I attached the new patch version.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Sun, Jan 18, 2009 at 9:51 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> I found this bug report : http://trac.wxwidgets.org/ticket/4841

Does the suggested fix work?

>> - The form underneath the AUI controls seems to have some drawing
>> issues on XP. The refresh bar is just about unusable for example.
>>
>
> Yeah. Even with this new patch, I still have this issue and no idea how
> to fix this. I tried the OnSize and OnEraseBackground methods, but that
> didn't fix it.

You might try sticking a panel on the form and then put the aui
notebook on that.

If we can't fix the GTK/wxAUI_NB_TAB_SPLIT bug it all seems a bit
academic though. That feature is what makes this worthwhile (the
ability to view multiple tabs on the status dialogue simultaneously -
for those that haven't been following closely :-p). I suppose we could
enable the feature just on Windows & Mac, but well, urgh...

Did you have any thoughts on the idea of changing the log viewer to a
text control instead of the listctrl?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Sun, Jan 18, 2009 at 9:51 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> I found this bug report : http://trac.wxwidgets.org/ticket/4841
>
> Does the suggested fix work?
>
>>> - The form underneath the AUI controls seems to have some drawing
>>> issues on XP. The refresh bar is just about unusable for example.
>>>
>> Yeah. Even with this new patch, I still have this issue and no idea how
>> to fix this. I tried the OnSize and OnEraseBackground methods, but that
>> didn't fix it.
>
> You might try sticking a panel on the form and then put the aui
> notebook on that.
>

They are now on a toolbar and I still have the issue. I probably did
something wrong, I'll check this today.

> If we can't fix the GTK/wxAUI_NB_TAB_SPLIT bug it all seems a bit
> academic though. That feature is what makes this worthwhile (the
> ability to view multiple tabs on the status dialogue simultaneously -
> for those that haven't been following closely :-p). I suppose we could
> enable the feature just on Windows & Mac, but well, urgh...
>

The bug report I found is not really my issue. The fix proposed does not
work, and I also have the issue with XFCE/Gnome.

I don't think the issue is in our code because the aui sample from the
wxWidgets source tarball behave exactly the same (ie it crashes).

I personaly would prefer to commit the patch (when the toolbar issue is
fixed). Of course, the wxAUI_NB_TAB_SPLIT will be disable on Linux.
Remember we also have this issue on the other pgadmin's Window :
frmMain, frmQuery, frmEditGrid. They all crash if I try to move a pane.

> Did you have any thoughts on the idea of changing the log viewer to a
> text control instead of the listctrl?
>

I didn't try but it should work. We'll need to trash the columns code,
but it won't be a hard work.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Wed, Jan 21, 2009 at 9:53 AM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> They are now on a toolbar and I still have the issue. I probably did
> something wrong, I'll check this today.

OK. I wonder if it would make some more sense to have separate
controls for each tab anyway, and just put them directly on each tab.
It might be preferable to be able to control the refresh rate of each
pane individually.

>> If we can't fix the GTK/wxAUI_NB_TAB_SPLIT bug it all seems a bit
>> academic though. That feature is what makes this worthwhile (the
>> ability to view multiple tabs on the status dialogue simultaneously -
>> for those that haven't been following closely :-p). I suppose we could
>> enable the feature just on Windows & Mac, but well, urgh...
>>
>
> The bug report I found is not really my issue. The fix proposed does not
> work, and I also have the issue with XFCE/Gnome.

OK.

> I don't think the issue is in our code because the aui sample from the
> wxWidgets source tarball behave exactly the same (ie it crashes).

Yeah.

> I personaly would prefer to commit the patch (when the toolbar issue is
> fixed). Of course, the wxAUI_NB_TAB_SPLIT will be disable on Linux.
> Remember we also have this issue on the other pgadmin's Window :
> frmMain, frmQuery, frmEditGrid. They all crash if I try to move a pane.

Except we're not proposing to add that functionality to those forms
(in fact, frmEditGrid doesn't even have a tabset). doing so would be a
lot of work, and I think it's only worth the effort perhaps on
frmMain.

>> Did you have any thoughts on the idea of changing the log viewer to a
>> text control instead of the listctrl?
>>
>
> I didn't try but it should work. We'll need to trash the columns code,
> but it won't be a hard work.

Cool.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Wed, Jan 21, 2009 at 9:53 AM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> They are now on a toolbar and I still have the issue. I probably did
>> something wrong, I'll check this today.
>
> OK. I wonder if it would make some more sense to have separate
> controls for each tab anyway, and just put them directly on each tab.
> It might be preferable to be able to control the refresh rate of each
> pane individually.
>

The patch attached adds the control on the activity status tab. If the
UI glitch the previous patch had doesn't appear with this one on
Windows, it could be interesting to add them on the three remaining
ones. Unfortunately, I can't check this this week (I won't have access
to a Windows PC till next week).

Can someone check this?

> [...]
>> I personaly would prefer to commit the patch (when the toolbar issue is
>> fixed). Of course, the wxAUI_NB_TAB_SPLIT will be disable on Linux.
>> Remember we also have this issue on the other pgadmin's Window :
>> frmMain, frmQuery, frmEditGrid. They all crash if I try to move a pane.
>
> Except we're not proposing to add that functionality to those forms
> (in fact, frmEditGrid doesn't even have a tabset). doing so would be a
> lot of work, and I think it's only worth the effort perhaps on
> frmMain.
>
>>> Did you have any thoughts on the idea of changing the log viewer to a
>>> text control instead of the listctrl?
>>>
>> I didn't try but it should work. We'll need to trash the columns code,
>> but it won't be a hard work.
>
> Cool.
>

So, what's next? would the patch be rejected if wxAUI_NB_TAB_SPLIT is
disabled? should I still work on this patch?

On my todo list, I have now:

* add copy of the query running on the selected backend (button before
Cancel Query)

* activity view : color coding
* new configuration tab (Server Status) : query time limit
* color activity view
green OK, orange running but query time past a configurable limit,
blue idle, red waiting for a lock
* menu view : item to (dis)allow color coding

* select a backend process and display just the locks it's holding, and
the one it's waiting on

* enable change of connection server

* add filter and sort for each report

* Add a dockable toolbar to enable changing connection server, timer
slider, filter and sort options.

* Add graph reports for different statistics views, something like
the Gnome System Monitor but specific to PostgreSQL (transaction
rate, read/hit ratio, number of connections, bgwriter stats come to
mind)

Comments?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Tue, Jan 27, 2009 at 2:30 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> The patch attached adds the control on the activity status tab. If the
> UI glitch the previous patch had doesn't appear with this one on
> Windows, it could be interesting to add them on the three remaining
> ones. Unfortunately, I can't check this this week (I won't have access
> to a Windows PC till next week).
>
> Can someone check this?

Looks good on Windows :) I see the buttons on each tab, but the slider
just on one which is what I guess you mean be 'add them on the three
remaining ones'.

> So, what's next? would the patch be rejected if wxAUI_NB_TAB_SPLIT is
> disabled? should I still work on this patch?

Well, if it's rejected it'll be by mutual agreement. I like the
feature enough that I think it probably should be enabled on Windows
and Mac, and disabled on GTK until the wx folks can fix things. It
would be worth logging/updating the bug with them.

> On my todo list, I have now:
>
> * add copy of the query running on the selected backend (button before
> Cancel Query)
>
> * activity view : color coding
> * new configuration tab (Server Status) : query time limit
> * color activity view
> green OK, orange running but query time past a configurable limit,
> blue idle, red waiting for a lock
> * menu view : item to (dis)allow color coding
>
> * select a backend process and display just the locks it's holding, and
> the one it's waiting on
>
> * enable change of connection server
>
> * add filter and sort for each report
>
> * Add a dockable toolbar to enable changing connection server, timer
> slider, filter and sort options.
>
> * Add graph reports for different statistics views, something like
> the Gnome System Monitor but specific to PostgreSQL (transaction
> rate, read/hit ratio, number of connections, bgwriter stats come to
> mind)
>
> Comments?

What is your timetable for this lot? I'm expecting us to freeze within
a few weeks. Tom wants us to get PG more or less ready for beta in two
(I think it'll take a little longer but...). We need to concentrate on
getting the changes in and complete one or two at a time until we run
out of time I guess.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Tue, Jan 27, 2009 at 2:30 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> The patch attached adds the control on the activity status tab. If the
>> UI glitch the previous patch had doesn't appear with this one on
>> Windows, it could be interesting to add them on the three remaining
>> ones. Unfortunately, I can't check this this week (I won't have access
>> to a Windows PC till next week).
>>
>> Can someone check this?
>
> Looks good on Windows :)

Great news.

> I see the buttons on each tab, but the slider
> just on one which is what I guess you mean be 'add them on the three
> remaining ones'.
>

Yeah, that's it. I'll do this right now.

>> So, what's next? would the patch be rejected if wxAUI_NB_TAB_SPLIT is
>> disabled? should I still work on this patch?
>
> Well, if it's rejected it'll be by mutual agreement. I like the
> feature enough that I think it probably should be enabled on Windows
> and Mac, and disabled on GTK until the wx folks can fix things.

OK. Good to hear :) I thought you really disliked the fact it could be
disabled on a plateforms and enabled on other ones. Dislike enough to
reject it.

> It would be worth logging/updating the bug with them.
>

Yes, I want to do that. I even thought I would work on that. But, with
the small delay we have, I prefer working on pgAdmin's code first.

>> On my todo list, I have now:
>>
>> * add copy of the query running on the selected backend (button before
>> Cancel Query)
>>
>> * activity view : color coding
>> * new configuration tab (Server Status) : query time limit
>> * color activity view
>> green OK, orange running but query time past a configurable limit,
>> blue idle, red waiting for a lock
>> * menu view : item to (dis)allow color coding
>>
>> * select a backend process and display just the locks it's holding, and
>> the one it's waiting on
>>
>> * enable change of connection server
>>
>> * add filter and sort for each report
>>
>> * Add a dockable toolbar to enable changing connection server, timer
>> slider, filter and sort options.
>>
>> * Add graph reports for different statistics views, something like
>> the Gnome System Monitor but specific to PostgreSQL (transaction
>> rate, read/hit ratio, number of connections, bgwriter stats come to
>> mind)
>>
>> Comments?
>
> What is your timetable for this lot? I'm expecting us to freeze within
> a few weeks. Tom wants us to get PG more or less ready for beta in two

That was also my idea after reading the whole thread on 8.4 release
planning.

> (I think it'll take a little longer but...).

Yes, probably.

> We need to concentrate on
> getting the changes in and complete one or two at a time until we run
> out of time I guess.
>

I'm OK for this.

If I have two weeks, I won't be able to do all of this TODO list. My
guess is that I can do at least the three first items (copy query (1
day), color coding of the activity view (three days), and the view of a
selected backend's locks (two days)). Next week will probably be quiet
for me, so I'll be able to do some pgAdmin's hacking. And I have the end
of this week too.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> On Tue, Jan 27, 2009 at 2:30 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>>> The patch attached adds the control on the activity status tab. If the
>>> UI glitch the previous patch had doesn't appear with this one on
>>> Windows, it could be interesting to add them on the three remaining
>>> ones. Unfortunately, I can't check this this week (I won't have access
>>> to a Windows PC till next week).
>>>
>>> Can someone check this?
>> Looks good on Windows :)
>
> Great news.
>
>> I see the buttons on each tab, but the slider
>> just on one which is what I guess you mean be 'add them on the three
>> remaining ones'.
>>
>
> Yeah, that's it. I'll do this right now.
>

OK, I have now four sliders, four preferences... but only one timer.
It's probably too late for me to think about it. Anyways, four timers
seem better. It would allow us to refresh each visible tab if one splits
the notebook (windows and mac only right now).

Any better ideas?

And is there something missing in this new patch?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Tue, Jan 27, 2009 at 11:23 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> OK, I have now four sliders, four preferences... but only one timer.
> It's probably too late for me to think about it. Anyways, four timers
> seem better. It would allow us to refresh each visible tab if one splits
> the notebook (windows and mac only right now).
>
> Any better ideas?

Nope - I think 4 timers is essential.

> And is there something missing in this new patch?

Do you mind if I review it again when the timer changes are done? I'm
pretty busy preparing for 8.3.6 this week, as well as doing some
urgent infrastructure work.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Tue, Jan 27, 2009 at 11:23 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> OK, I have now four sliders, four preferences... but only one timer.
>> It's probably too late for me to think about it. Anyways, four timers
>> seem better. It would allow us to refresh each visible tab if one splits
>> the notebook (windows and mac only right now).
>>
>> Any better ideas?
>
> Nope - I think 4 timers is essential.
>

I didn't get any better ideas myself... so I'll go with four timers.

>> And is there something missing in this new patch?
>
> Do you mind if I review it again when the timer changes are done? I'm
> pretty busy preparing for 8.3.6 this week, as well as doing some
> urgent infrastructure work.
>

No problem at all. I don't want to waste your time with this patch
version. I really prefer you check next one.

Thanks.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> On Tue, Jan 27, 2009 at 11:23 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>>
>>> OK, I have now four sliders, four preferences... but only one timer.
>>> It's probably too late for me to think about it. Anyways, four timers
>>> seem better. It would allow us to refresh each visible tab if one splits
>>> the notebook (windows and mac only right now).
>>>
>>> Any better ideas?
>> Nope - I think 4 timers is essential.
>>
>
> I didn't get any better ideas myself... so I'll go with four timers.
>
>>> And is there something missing in this new patch?
>> Do you mind if I review it again when the timer changes are done? I'm
>> pretty busy preparing for 8.3.6 this week, as well as doing some
>> urgent infrastructure work.
>>
>
> No problem at all. I don't want to waste your time with this patch
> version. I really prefer you check next one.
>

So here is the next one. I fixed a few issues, added some more comments,
moved some code to get something easy to read, and used the four timers.
It seems to be working right on my laptop. I did a quick check on
Windows and it was working all right too.

I did get a segmentation fault once when doing this: I launched pgAdmin,
opened a query window and the server status, then closed the query
window and... kaboom. Segmentation fault. I'm not able to reproduce
this. Perhaps I forgot one step.

So, please, check this patch carefully. I'll continue to test it.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Jan 29, 2009 at 10:40 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> So here is the next one. I fixed a few issues, added some more comments,
> moved some code to get something easy to read, and used the four timers.
> It seems to be working right on my laptop. I did a quick check on
> Windows and it was working all right too.

Hi,

There are a few issues immediately apparent:

- There's debug code in checkConnection

- MNU_REFRESH/OnRefresh() seem redundant now.

- The default size of the dialog is tall and thin - if I so much as
touch the resize corner it doubles in width to a more sensible size.
Observed on Windows.

- You need a lock to prevent the four refresh functions running
simultaneously and doing nasty things to the connection. You should
probably use a wxCriticalSectionLocker.

- It looks like we still refresh each tab even if it's not visible (or
even closed). We should check that to minimise the database traffic.

- If a tab is closed and re-opened (from the View menu), it returns to
it's default position on the tabset, not to whereever it previously
was.

- The tab layout is not remembered if the window is closed and reopened.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Hi,

Dave Page a écrit :
> On Thu, Jan 29, 2009 at 10:40 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> So here is the next one. I fixed a few issues, added some more comments,
>> moved some code to get something easy to read, and used the four timers.
>> It seems to be working right on my laptop. I did a quick check on
>> Windows and it was working all right too.
>
> Hi,
>
> There are a few issues immediately apparent:
>
> - There's debug code in checkConnection
>

Oops, sorry about this. Fixed.

> - MNU_REFRESH/OnRefresh() seem redundant now.
>

I removed the MNU_REFRESH but kept OnRefresh(). frmStatus still uses it
(and a lot).

> - The default size of the dialog is tall and thin - if I so much as
> touch the resize corner it doubles in width to a more sensible size.
> Observed on Windows.
>

Strange behavior. It didn't occur on GTK. Anyways, fixed.

> - You need a lock to prevent the four refresh functions running
> simultaneously and doing nasty things to the connection. You should
> probably use a wxCriticalSectionLocker.
>

wxCriticalSectionLocker added.

> - It looks like we still refresh each tab even if it's not visible (or
> even closed). We should check that to minimise the database traffic.
>

Fixed.

> - If a tab is closed and re-opened (from the View menu), it returns to
> it's default position on the tabset, not to whereever it previously
> was.
>
> - The tab layout is not remembered if the window is closed and reopened.
>

The last two ones aren't done on purpose.

On the first one, it doesn't return on its default position but on the
last position. I don't think it's really interesting and it will be
difficult to do. For example, how do we handle the case when we have
multiple notebooks?

On the second one, if we try to remember the tab layout (not difficult),
do we also need to remember the pane layout? much more difficult.

I would prefer to spend the rest of the time left with the other todo items.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Hi,

On Tue, Feb 3, 2009 at 5:07 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

>> - If a tab is closed and re-opened (from the View menu), it returns to
>> it's default position on the tabset, not to whereever it previously
>> was.
>>
>> - The tab layout is not remembered if the window is closed and reopened.
>>
>
> The last two ones aren't done on purpose.
>
> On the first one, it doesn't return on its default position but on the
> last position. I don't think it's really interesting and it will be
> difficult to do. For example, how do we handle the case when we have
> multiple notebooks?

It does seem to go to the default position here. If I close the a tab,
then no matter where it was, it returns to the end on the main tabset.

Thinking back to the original AUI code, iirc, it's the manager that
actually remembers where everything is, so I can see it may be hard to
implement for tabs. I'm don't think it's that important.

> On the second one, if we try to remember the tab layout (not difficult),
> do we also need to remember the pane layout? much more difficult.

We should certainly record and restore the perspective (as we seem to
be), but it doesn't appear to be doing anything. I would expect the
layout to be restored, so if I run the tool every morning with a
preferred layout (say three tabs, two at the top, one below, all
visible at once) I don't have to re-layout everything again. Consider
Visual Studio, in which the layout is retained but would be quite
frustrating if it wasn't.

I also noticed another couple of issues:

- If you place 2 tabs side-by-side, the Refresh button and slider can
become easily obscured, but it is not obvious that anything has been
lost. I wonder if we can wrap those controls, or otherwise improve the
problem somehow. Any ideas?

- The tab names should match the text on the View menu. Specifically,
Status vs. Activity

- After closing a tab, the 'Default View' option can open 2 copies of
some tabs. It's quite easy to get 6 or more tabs and then have it
crash.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Hi,

Dave Page a écrit :
> [...]
> On Tue, Feb 3, 2009 at 5:07 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>>> - If a tab is closed and re-opened (from the View menu), it returns to
>>> it's default position on the tabset, not to whereever it previously
>>> was.
>>>
>>> - The tab layout is not remembered if the window is closed and reopened.
>>>
>> The last two ones aren't done on purpose.
>>
>> On the first one, it doesn't return on its default position but on the
>> last position. I don't think it's really interesting and it will be
>> difficult to do. For example, how do we handle the case when we have
>> multiple notebooks?
>
> It does seem to go to the default position here. If I close the a tab,
> then no matter where it was, it returns to the end on the main tabset.
>

Yes, I wanted to say this :) sorry, my explanations weren't good.

> Thinking back to the original AUI code, iirc, it's the manager that
> actually remembers where everything is, so I can see it may be hard to
> implement for tabs. I'm don't think it's that important.
>

If there's only one notebook, it's easy to remember one tab's location
when it's closed. But if there are many split part of a notebook, I
can't add a tab to a specific split part.

>> On the second one, if we try to remember the tab layout (not difficult),
>> do we also need to remember the pane layout? much more difficult.
>
> We should certainly record and restore the perspective (as we seem to
> be), but it doesn't appear to be doing anything. I would expect the
> layout to be restored, so if I run the tool every morning with a
> preferred layout (say three tabs, two at the top, one below, all
> visible at once) I don't have to re-layout everything again. Consider
> Visual Studio, in which the layout is retained but would be quite
> frustrating if it wasn't.
>

I understand, and I agree with you. The application can split a
wxAuiNotebook (with the Split() method). Unfortunately, we can't change
its size, position, and contents. I'm afraid this widget won't allow us
to do what we want.

The only way left is to use multiple panes with one wxAuiNoteBook on
each pane... and disallow the split stuff.

> I also noticed another couple of issues:
>
> - If you place 2 tabs side-by-side, the Refresh button and slider can
> become easily obscured, but it is not obvious that anything has been
> lost. I wonder if we can wrap those controls, or otherwise improve the
> problem somehow. Any ideas?
>

They're already wrapped in a wxFlexGridSizer. I don't think we can do
much with since. Perhaps a toolbar for each tab? but I don't really like
this idea.

> - The tab names should match the text on the View menu. Specifically,
> Status vs. Activity
>

Fixed.

> - After closing a tab, the 'Default View' option can open 2 copies of
> some tabs. It's quite easy to get 6 or more tabs and then have it
> crash.
>

Fixed.

New patch attached.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> [...]
>>> On the second one, if we try to remember the tab layout (not difficult),
>>> do we also need to remember the pane layout? much more difficult.
>> We should certainly record and restore the perspective (as we seem to
>> be), but it doesn't appear to be doing anything. I would expect the
>> layout to be restored, so if I run the tool every morning with a
>> preferred layout (say three tabs, two at the top, one below, all
>> visible at once) I don't have to re-layout everything again. Consider
>> Visual Studio, in which the layout is retained but would be quite
>> frustrating if it wasn't.
>>
>
> I understand, and I agree with you. The application can split a
> wxAuiNotebook (with the Split() method). Unfortunately, we can't change
> its size, position, and contents. I'm afraid this widget won't allow us
> to do what we want.
>
> The only way left is to use multiple panes with one wxAuiNoteBook on
> each pane... and disallow the split stuff.
>

Comments? do I try to use multiple panes?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Sorry - things got hectic again.

On Tue, Feb 10, 2009 at 8:33 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> I understand, and I agree with you. The application can split a
> wxAuiNotebook (with the Split() method). Unfortunately, we can't change
> its size, position, and contents. I'm afraid this widget won't allow us
> to do what we want.
>
> The only way left is to use multiple panes with one wxAuiNoteBook on
> each pane... and disallow the split stuff.

So essentially what we're saying is that we either cannot save the
layout, or we have to disable the only feature that makes it
worthwhile using wxAuiNotebook instead of wxNotebook? Bummer.

What about not using a notebook at all and just using 4 panels, like
we do on other forms? We can have toggle button on a button bar as
well as menu options to turn each pane on and off. that will allow
flexible layout within the form, or for sections to be un-docked into
their own Windows.

> They're already wrapped in a wxFlexGridSizer. I don't think we can do
> much with since. Perhaps a toolbar for each tab? but I don't really like
> this idea.

What about replacing the slider and text control with a combo box that
offers Don't refresh, ! second, 5 seconds, 10 seconds, 30 seconds, 1
minute, 5 minutes, 10 minutes, 30 minutes, 1 hour. That will be a
fixed size (and much smaller than the slider), and should be small
enough to fit a decent four-pane layout on an 800x600 window.

Thoughts?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> Sorry - things got hectic again.
>

Yeah, don't worry about this. I sent this mail because of the small
delay we still have.

> On Tue, Feb 10, 2009 at 8:33 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> I understand, and I agree with you. The application can split a
>> wxAuiNotebook (with the Split() method). Unfortunately, we can't change
>> its size, position, and contents. I'm afraid this widget won't allow us
>> to do what we want.
>>
>> The only way left is to use multiple panes with one wxAuiNoteBook on
>> each pane... and disallow the split stuff.
>
> So essentially what we're saying is that we either cannot save the
> layout, or we have to disable the only feature that makes it
> worthwhile using wxAuiNotebook instead of wxNotebook? Bummer.
>
> What about not using a notebook at all and just using 4 panels, like
> we do on other forms? We can have toggle button on a button bar as
> well as menu options to turn each pane on and off. that will allow
> flexible layout within the form, or for sections to be un-docked into
> their own Windows.
>

So, we keep the wxAuiManager and add panes. I'm not sure if the result
will be really interesting. But I don't think it is much work to do.
I'll work on it right now.

>> They're already wrapped in a wxFlexGridSizer. I don't think we can do
>> much with since. Perhaps a toolbar for each tab? but I don't really like
>> this idea.
>
> What about replacing the slider and text control with a combo box that
> offers Don't refresh, ! second, 5 seconds, 10 seconds, 30 seconds, 1
> minute, 5 minutes, 10 minutes, 30 minutes, 1 hour. That will be a
> fixed size (and much smaller than the slider), and should be small
> enough to fit a decent four-pane layout on an 800x600 window.
>

OK. If people wants more delay options, we can let the user customize
this with options in the options dialog.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Feb 12, 2009 at 4:26 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> So, we keep the wxAuiManager and add panes. I'm not sure if the result
> will be really interesting. But I don't think it is much work to do.
> I'll work on it right now.

Thats what I figured. Just stick everything on a pane instead of a
notebook page, and away it goes.

>>> They're already wrapped in a wxFlexGridSizer. I don't think we can do
>>> much with since. Perhaps a toolbar for each tab? but I don't really like
>>> this idea.
>>
>> What about replacing the slider and text control with a combo box that
>> offers Don't refresh, ! second, 5 seconds, 10 seconds, 30 seconds, 1
>> minute, 5 minutes, 10 minutes, 30 minutes, 1 hour. That will be a
>> fixed size (and much smaller than the slider), and should be small
>> enough to fit a decent four-pane layout on an 800x600 window.
>>
>
> OK. If people wants more delay options, we can let the user customize
> this with options in the options dialog.

I don't think there's any need though, do you?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Thu, Feb 12, 2009 at 4:26 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> So, we keep the wxAuiManager and add panes. I'm not sure if the result
>> will be really interesting. But I don't think it is much work to do.
>> I'll work on it right now.
>
> Thats what I figured. Just stick everything on a pane instead of a
> notebook page, and away it goes.
>

The patch attached takes care of this. It's not really finished but it
gives an idea of what could be done. There's also the combo stuff. I
would love to get comments on this, to see if I go further.

>>>> They're already wrapped in a wxFlexGridSizer. I don't think we can do
>>>> much with since. Perhaps a toolbar for each tab? but I don't really like
>>>> this idea.
>>> What about replacing the slider and text control with a combo box that
>>> offers Don't refresh, ! second, 5 seconds, 10 seconds, 30 seconds, 1
>>> minute, 5 minutes, 10 minutes, 30 minutes, 1 hour. That will be a
>>> fixed size (and much smaller than the slider), and should be small
>>> enough to fit a decent four-pane layout on an 800x600 window.
>>>
>> OK. If people wants more delay options, we can let the user customize
>> this with options in the options dialog.
>
> I don't think there's any need though, do you?
>

Not now, but perhaps in a future release.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Feb 12, 2009 at 11:09 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> The patch attached takes care of this. It's not really finished but it
> gives an idea of what could be done. There's also the combo stuff. I
> would love to get comments on this, to see if I go further.

I immediately got an assertion (well, four):

---------------------------
wxWidgets Debug Alert
---------------------------
..\..\src\msw\registry.cpp(961): assert "!IsNumericValue(szValue)"
failed in wxRegKey::QueryValue(): Type mismatch in
wxRegKey::QueryValue().

Call stack:
[00] wxGUIAppTraitsBase::ShowAssertDialog
c:\pgbuild\wxwidgets\src\common\appcmn.cpp:635
[01] ShowAssertDialog
c:\pgbuild\wxwidgets\src\common\appbase.cpp:836
[02] wxAppConsole::OnAssertFailure
c:\pgbuild\wxwidgets\src\common\appbase.cpp:445
[03] wxOnAssert
c:\pgbuild\wxwidgets\src\common\appbase.cpp:713
[04] wxRegKey::QueryValue
c:\pgbuild\wxwidgets\src\msw\registry.cpp:961
[05] wxRegKey::QueryValue
c:\pgbuild\wxwidgets\include\wx\msw\registry.h:167
[06] TryGetValue
c:\pgbuild\wxwidgets\src\msw\regconf.cpp:48
[07] wxRegConfig::DoReadString
c:\pgbuild\wxwidgets\src\msw\regconf.cpp:578
[08] wxConfigBase::Read
c:\pgbuild\wxwidgets\src\common\config.cpp:143
[09] sysSettings::Read
c:\pgbuild\buildtrees\pgadmin3\pgadmin\utils\syssettings.cpp:266
[10] frmStatus::AddStatusPage
c:\pgbuild\buildtrees\pgadmin3\pgadmin\frm\frmstatus.cpp:379
[11] frmStatus::frmStatus
c:\pgbuild\buildtrees\pgadmin3\pgadmin\frm\frmstatus.cpp:220
[12] serverStatusFactory::StartDialog
c:\pgbuild\buildtrees\pgadmin3\pgadmin\frm\frmstatus.cpp:1824
[13] frmMain::OnAction
c:\pgbuild\buildtrees\pgadmin3\pgadmin\frm\events.cpp:147
[14] wxAppConsole::HandleEvent
c:\pgbuild\wxwidgets\src\common\appbase.cpp:323
[15] wxEvtHandler::ProcessEventIfMatches
c:\pgbuild\wxwidgets\src\common\event.cpp:1233
[16] wxEvtHandler::SearchDynamicEventTable
c:\pgbuild\wxwidgets\src\common\event.cpp:1413
[17] wxEvtHandler::ProcessEvent
c:\pgbuild\wxwidgets\src\common\event.cpp:1289
[18] wxEvtHandler::ProcessEvent
c:\pgbuild\wxwidgets\src\common\event.cpp:1300
[19] wxFrameBase::ProcessCommand
c:\pgbuild\wxwidgets\src\common\framecmn.cpp:224
[20] wxFrame::HandleCommand
c:\pgbuild\wxwidgets\src\msw\frame.cpp:978
Do you want to stop the program?
You can also choose [Cancel] to suppress further warnings.
---------------------------
Yes No Cancel
---------------------------

I suspect, that was because of the existing refresh rate(s) being
stored in the registry.

Aside from that, I quite like this layout :-) A few thoughts, in no
particular order:

- The layout is not being restored when the window is re-opened.
Should be easy to do - the code already exists in other forms.

- The main pane on the right should perhaps be the log pane, as thats
the one that will usually have lots of info in it, even on a developer
system with no other users.

- Can we move the per-pane controls to a single toolbar at the top of
the form, and have that display only the controls for the active pane?
That would be a lot tidier I think, and would largely solve the
pane-too-small-tools-truncated-or-obscured problem. There is a flag to
the AUI manager to have it highlight the top bar of the active pane
iirc, so it should be obvious which pane a user would be modifying.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Thu, Feb 12, 2009 at 11:09 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> The patch attached takes care of this. It's not really finished but it
>> gives an idea of what could be done. There's also the combo stuff. I
>> would love to get comments on this, to see if I go further.
>
> I immediately got an assertion (well, four):
> [...]
> I suspect, that was because of the existing refresh rate(s) being
> stored in the registry.
>

Sorry, I should have warned you...

> Aside from that, I quite like this layout :-) A few thoughts, in no
> particular order:
>
> - The layout is not being restored when the window is re-opened.
> Should be easy to do - the code already exists in other forms.
>

The part of this code is commented on the patch. I should have better
looked at the patch before sending it.

> - The main pane on the right should perhaps be the log pane, as thats
> the one that will usually have lots of info in it, even on a developer
> system with no other users.
>

Done.

> - Can we move the per-pane controls to a single toolbar at the top of
> the form, and have that display only the controls for the active pane?
> That would be a lot tidier I think, and would largely solve the
> pane-too-small-tools-truncated-or-obscured problem. There is a flag to
> the AUI manager to have it highlight the top bar of the active pane
> iirc, so it should be obvious which pane a user would be modifying.
>

I'll look at this.

Thanks for your comments, and sorry for the assertion.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> On Thu, Feb 12, 2009 at 11:09 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
> [...]
>> - Can we move the per-pane controls to a single toolbar at the top of
>> the form, and have that display only the controls for the active pane?
>> That would be a lot tidier I think, and would largely solve the
>> pane-too-small-tools-truncated-or-obscured problem. There is a flag to
>> the AUI manager to have it highlight the top bar of the active pane
>> iirc, so it should be obvious which pane a user would be modifying.
>>
>
> I'll look at this.
>

Done. It was much harder than I initially thought but I really like this
one. I checked on each platform and it behaves nicely on each.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Wed, Feb 18, 2009 at 11:02 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> Done. It was much harder than I initially thought but I really like this
> one. I checked on each platform and it behaves nicely on each.

Yeah, that looks much nicer. I tried it on Windows and Mac, and came
across the following issues. Note that I've attached a couple of
screenshots of the default layouts (ie. having removed all frmStatus
related settings from the ini file/registry):

- The Refresh rate combo box is top-justified on Mac. Why is it on a
separate toolbar though?

- The terminate backend button icon should at least be a different
colour from the cancel button.

- If I undock a pane on Mac, the default size is about 2 pixels wide
by 20 high (thats a guess). Makes it pretty hard to see or resize :-p

- If I set a different refresh time for each pane, I would expect to
see the value change in the combo box when I select different panes.

- Should we use wxAUI_MGR_ALLOW_ACTIVE_PANE when we setup the manager?
That will display the title bar of the active pane in a different
colour, but it's not entirely tied to whether or not the listview is
active or not. You can see this in the Mac screenshot.

- The timer defaults should probably be more like 10 seconds than 1.

- There should be a view option for the toolbar.

- The default layout does not work with the default size. See the screenshots.

We're getting there though :-)

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Wed, Feb 18, 2009 at 11:02 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> Done. It was much harder than I initially thought but I really like this
>> one. I checked on each platform and it behaves nicely on each.
>
> Yeah, that looks much nicer. I tried it on Windows and Mac, and came
> across the following issues. Note that I've attached a couple of
> screenshots of the default layouts (ie. having removed all frmStatus
> related settings from the ini file/registry):
>
> - The Refresh rate combo box is top-justified on Mac. Why is it on a
> separate toolbar though?
>

I just copied and pasted the frmQuery example :)

The new patch uses only one toolbar. The combobox is no more
top-justified on Mac.

> - The terminate backend button icon should at least be a different
> colour from the cancel button.
>

Hmmm, I'm not really good at drawing. Anyway, I added some red in it.

> - If I undock a pane on Mac, the default size is about 2 pixels wide
> by 20 high (thats a guess). Makes it pretty hard to see or resize :-p
>

I'm a bit lost here. I have some issues on Mac, but I don't know how to
delete saved preferences. On GTK, it's simple. I just need to edit the
~/.pgadmin3 file. On Windows, it's a bit more difficult but with
regedit, one can do it. But I don't find a way to do this on Mac. Can
you shared your precious knowledge ? :)

> - If I set a different refresh time for each pane, I would expect to
> see the value change in the combo box when I select different panes.
>

Well, it should. And it does when you click on another line than the one
already selected. EVT_LIST_ITEM_SELECTED doesn't get fired when the user
clicks on the already selected row. And I don't find another "activate"
event. Do you one that I can use ?

> - Should we use wxAUI_MGR_ALLOW_ACTIVE_PANE when we setup the manager?
> That will display the title bar of the active pane in a different
> colour, but it's not entirely tied to whether or not the listview is
> active or not. You can see this in the Mac screenshot.
>

Done.

> - The timer defaults should probably be more like 10 seconds than 1.
>

Fixed.

> - There should be a view option for the toolbar.
>

Done too.

> - The default layout does not work with the default size. See the screenshots.
>

I changed them. Hope it will work better.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Tue, Feb 24, 2009 at 6:53 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

>> - If I undock a pane on Mac, the default size is about 2 pixels wide
>> by 20 high (thats a guess). Makes it pretty hard to see or resize :-p
>>
>
> I'm a bit lost here. I have some issues on Mac, but I don't know how to
> delete saved preferences. On GTK, it's simple. I just need to edit the
> ~/.pgadmin3 file. On Windows, it's a bit more difficult but with
> regedit, one can do it. But I don't find a way to do this on Mac. Can
> you shared your precious knowledge ? :)

Look in ~/Library/Preferences/pgadmin3\ Preferences

(yes, that is a space in there. Don't blame me :-p)

Looking at the patch now btw - will reply in more detail soon.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

OK, fresh message with test results from the v13 patch :-)

- terminate_backend.xpm is missing

- The default perspectives are still not good (at least on Windows).
The toolbar is still sized so that the refresh combo box doesn't fit -
actually I'm not sure if this isn't another bug. Using the 'Default
View' option didn't help. It got fixed when I removed all the existing
registry settings for frmStatus and opened the Window again. Related
to that, if I un-dock a pane on Windows, currently the default size is
approx 1cm^2, similar to the problem I noticed on the Mac :-)

- Can you please add dividers before the logfile combo box, and after
the rotate button to separate them from the other controls.

- We need an event that can capture changes to the active pane
properly (as you noted). I'll look at that in a minute and see what I
can figure out. The current behaviour is too unpredictable.

With regard to the perspective issues, here's how I normally deal with them:

- Delete all existing stored settings, and open the dialog.

- Size each pane appropriately, in the default-sized window.

- Undock each pane in turn, and size it appropriately. Redock when
sized. You can undock and redock again to check that the correct size
is retained

- Close the window, and grab the perspective setting from the
registry/config file, and copy to the source file.

- Repeat on each of the three platforms.

Nearly there :-)

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Forgot to mention - what happened about changing the log pane to use a
text control?

On Wed, Feb 25, 2009 at 11:22 AM, Dave Page <dpage(at)pgadmin(dot)org> wrote:
> OK, fresh message with test results from the v13 patch :-)
>
> - terminate_backend.xpm is missing
>
> - The default perspectives are still not good (at least on Windows).
> The toolbar is still sized so that the refresh combo box doesn't fit -
> actually I'm not sure if this isn't another bug. Using the 'Default
> View' option didn't help. It got fixed when I removed all the existing
> registry settings for frmStatus and opened the Window again. Related
> to that, if I un-dock a pane on Windows, currently the default size is
> approx 1cm^2, similar to the problem I noticed on the Mac :-)
>
> - Can you please add dividers before the logfile combo box, and after
> the rotate button to separate them from the other controls.
>
> - We need an event that can capture changes to the active pane
> properly (as you noted). I'll look at that in a minute and see what I
> can figure out. The current behaviour is too unpredictable.
>
>
> With regard to the perspective issues, here's how I normally deal with them:
>
> - Delete all existing stored settings, and open the dialog.
>
> - Size each pane appropriately, in the default-sized window.
>
> - Undock each pane in turn, and size it appropriately. Redock when
> sized. You can undock and redock again to check that the correct size
> is retained
>
> - Close the window, and grab the perspective setting from the
> registry/config file, and copy to the source file.
>
> - Repeat on each of the three platforms.
>
> Nearly there :-)
>
> --
> Dave Page
> EnterpriseDB UK:   http://www.enterprisedb.com
>

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Tue, Feb 24, 2009 at 6:53 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>>> - If I undock a pane on Mac, the default size is about 2 pixels wide
>>> by 20 high (thats a guess). Makes it pretty hard to see or resize :-p
>>>
>> I'm a bit lost here. I have some issues on Mac, but I don't know how to
>> delete saved preferences. On GTK, it's simple. I just need to edit the
>> ~/.pgadmin3 file. On Windows, it's a bit more difficult but with
>> regedit, one can do it. But I don't find a way to do this on Mac. Can
>> you shared your precious knowledge ? :)
>
> Look in ~/Library/Preferences/pgadmin3\ Preferences
>
> (yes, that is a space in there. Don't blame me :-p)
>
> Looking at the patch now btw - will reply in more detail soon.
>

Thanks, I'll get a look a this.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Dave Page a écrit :
> OK, fresh message with test results from the v13 patch :-)
>
> - terminate_backend.xpm is missing
>

Sorry about this. I'll send a new patch with this file.

> - The default perspectives are still not good (at least on Windows).

I didn't change it on Mac because I couldn't find a way to get the
recorded preference.

> The toolbar is still sized so that the refresh combo box doesn't fit -
> actually I'm not sure if this isn't another bug. Using the 'Default
> View' option didn't help. It got fixed when I removed all the existing
> registry settings for frmStatus and opened the Window again. Related
> to that, if I un-dock a pane on Windows, currently the default size is
> approx 1cm^2, similar to the problem I noticed on the Mac :-)
>

I always do my tests without old settings. Strange that the Default View
did nothing.

> - Can you please add dividers before the logfile combo box, and after
> the rotate button to separate them from the other controls.
>

No problem.

> - We need an event that can capture changes to the active pane
> properly (as you noted). I'll look at that in a minute and see what I
> can figure out. The current behaviour is too unpredictable.
>
>
> With regard to the perspective issues, here's how I normally deal with them:
>
> - Delete all existing stored settings, and open the dialog.
>
> - Size each pane appropriately, in the default-sized window.
>
> - Undock each pane in turn, and size it appropriately. Redock when
> sized. You can undock and redock again to check that the correct size
> is retained
>
> - Close the window, and grab the perspective setting from the
> registry/config file, and copy to the source file.
>
> - Repeat on each of the three platforms.
>

I already did that, but without the "undock" step. I'll do it again
before sending v14.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Dave Page a écrit :
> Forgot to mention - what happened about changing the log pane to use a
> text control?
>

It seems I forgot it. Sorry.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> OK, fresh message with test results from the v13 patch :-)
>>
>> - terminate_backend.xpm is missing
>>
>
> Sorry about this. I'll send a new patch with this file.
>

Done.

>> - The default perspectives are still not good (at least on Windows).
>
> I didn't change it on Mac because I couldn't find a way to get the
> recorded preference.
>

Your method works great.

>> The toolbar is still sized so that the refresh combo box doesn't fit -
>> actually I'm not sure if this isn't another bug. Using the 'Default
>> View' option didn't help. It got fixed when I removed all the existing
>> registry settings for frmStatus and opened the Window again. Related
>> to that, if I un-dock a pane on Windows, currently the default size is
>> approx 1cm^2, similar to the problem I noticed on the Mac :-)
>>
>
> I always do my tests without old settings. Strange that the Default View
> did nothing.
>

Fixed.

>> - Can you please add dividers before the logfile combo box, and after
>> the rotate button to separate them from the other controls.
>>
>
> No problem.
>

Done.

>> - We need an event that can capture changes to the active pane
>> properly (as you noted). I'll look at that in a minute and see what I
>> can figure out. The current behaviour is too unpredictable.
>>

Fixed with your "truly horrible kludge". The wxTimer is set to half a
second, and it doesn't enough on Mac. Maybe 400ms would be better.

>> With regard to the perspective issues, here's how I normally deal with them:
>>
>> - Delete all existing stored settings, and open the dialog.
>>
>> - Size each pane appropriately, in the default-sized window.
>>
>> - Undock each pane in turn, and size it appropriately. Redock when
>> sized. You can undock and redock again to check that the correct size
>> is retained
>>
>> - Close the window, and grab the perspective setting from the
>> registry/config file, and copy to the source file.
>>
>> - Repeat on each of the three platforms.
>>
>
> I already did that, but without the "undock" step. I'll do it again
> before sending v14.
>

Done. I finally understood what's in the perspective. I finally found
editing the string manually was a better way.

I still didn't change the listview with a textbox for the Logfile pane.
It's a big work and I would prefer to finish this one before doing the
change.

See the attached patch.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

2009/2/25 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
> (...)
> See the attached patch.

Hi Guillaume,

OS: Ubuntu 8.10 Linux guedes-laptop 2.6.27-11-generic #1 SMP Thu Jan
29 19:24:39 UTC 2009 i686 GNU/Linux
SVN Revision: 7611
Patch: serverstatus_p1_v14d.patch.bz2

Problem 1: When status window opens an error appears
(serverstatus_p1_v14d.patch.error01.png)

Problem 2: I don't have log_filename in format
'postgresql-%Y-%m-%d_%H%M%S.log', so a second window appears and saw
that log_filename must equal 'postgresql-%Y-%m-%d_%H%M%S.log'
(serverstatus_p1_v14d.patch.error01.png), so I stay that windows
there, switch to my ssh shell terminal, change the parameter on
postgresql.conf, reload pg, come back to pgadmin and ouch! it is
frozen!(serverstatus_p1_v14d.patch.error03.png). The same occurs
without touch the conf, just waiting with that error window opened...
":(

GDB don't shows problems...

I hope this help in something...

Dickson S. Guedes
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

Hi Dave,

The last mail i've sent to you and Guillaume has the wrong images
because it contains some personal information that couldn't be public.

Is there a way to don't turn it public if it was bounced yet? I see my
mistake only now ":(

Thanks.

2009/2/26 Dickson S. Guedes <listas(at)guedesoft(dot)net>:
> 2009/2/25 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
>> (...)
>> See the attached patch.
>
> Hi Guillaume,
>
> OS: Ubuntu 8.10 Linux guedes-laptop 2.6.27-11-generic #1 SMP Thu Jan
> 29 19:24:39 UTC 2009 i686 GNU/Linux
> SVN Revision: 7611
> Patch: serverstatus_p1_v14d.patch.bz2
>
> Problem 1: When status window opens an error appears
> (serverstatus_p1_v14d.patch.error01.png)
>
> Problem 2: I don't have log_filename in format
> 'postgresql-%Y-%m-%d_%H%M%S.log', so a second window appears and saw
> that log_filename must equal 'postgresql-%Y-%m-%d_%H%M%S.log'
> (serverstatus_p1_v14d.patch.error01.png), so I stay that windows
> there, switch to my ssh shell terminal, change the parameter on
> postgresql.conf, reload pg, come back to pgadmin and ouch! it is
> frozen!(serverstatus_p1_v14d.patch.error03.png). The same occurs
> without touch the conf, just waiting with that error window opened...
> ":(
>
> GDB don't shows problems...
>
> I hope this help in something...
>
> Dickson S. Guedes
> mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
> http://guedesoft.net - http://planeta.postgresql.org.br
>

--
Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

2009/2/25 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
> See the attached patch.

Hi Guillaume,

OS: Ubuntu 8.10 Linux guedes-laptop 2.6.27-11-generic #1 SMP Thu Jan
29 19:24:39 UTC 2009 i686
SVN Revision: 7611
Patch: serverstatus_p1_v14d.patch.bz2

Problem 1: When status window opens an error appears ([1])

Problem 2: I don't have log_filename in format
'postgresql-%Y-%m-%d_%H%M%S.log', so a second window appears and saw
that log_filename must be equal 'postgresql-%Y-%m-%d_%H%M%S.log'
([2]), so I stay that window there, switch to my ssh shell terminal,
change the parameter on postgresql.conf, reload pg, come back to
pgadmin and ouch! it is frozen!([3]). The same occurs without touch
the conf, just waiting with that error window opened... ":(

GDB don't shows problems...

I hope this help in something...

[1] http://guedesoft.net/images/serverstatus_p1_v14d.patch.error01.png
[2] http://guedesoft.net/images/serverstatus_p1_v14d.patch.error02.png
[3] http://guedesoft.net/images/serverstatus_p1_v14d.patch.error03.png

--
Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

Dickson S. Guedes a écrit :
> 2009/2/25 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
>> See the attached patch.
>
> Hi Guillaume,
>
> OS: Ubuntu 8.10 Linux guedes-laptop 2.6.27-11-generic #1 SMP Thu Jan
> 29 19:24:39 UTC 2009 i686
> SVN Revision: 7611
> Patch: serverstatus_p1_v14d.patch.bz2
>
> Problem 1: When status window opens an error appears ([1])
>

That's not really en error message. It's a debug message. But you're
right. I have to delete it. THanks for reminding me.

> Problem 2: I don't have log_filename in format
> 'postgresql-%Y-%m-%d_%H%M%S.log', so a second window appears and saw
> that log_filename must be equal 'postgresql-%Y-%m-%d_%H%M%S.log'
> ([2]), so I stay that window there, switch to my ssh shell terminal,
> change the parameter on postgresql.conf, reload pg, come back to
> pgadmin and ouch! it is frozen!([3]). The same occurs without touch
> the conf, just waiting with that error window opened... ":(
>

I thought I was the only one who has this message. I'll work on this.

> GDB don't shows problems...
>
> I hope this help in something...
>

Yes, greatly. I work on these two right now. Thanks for your review.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dickson S. Guedes a écrit :
>> 2009/2/25 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
>>> See the attached patch.
>> Hi Guillaume,
>>
>> OS: Ubuntu 8.10 Linux guedes-laptop 2.6.27-11-generic #1 SMP Thu Jan
>> 29 19:24:39 UTC 2009 i686
>> SVN Revision: 7611
>> Patch: serverstatus_p1_v14d.patch.bz2
>>
>> Problem 1: When status window opens an error appears ([1])
>>
>
> That's not really en error message. It's a debug message. But you're
> right. I have to delete it. THanks for reminding me.
>

Fixed in this new patch.

>> Problem 2: I don't have log_filename in format
>> 'postgresql-%Y-%m-%d_%H%M%S.log', so a second window appears and saw
>> that log_filename must be equal 'postgresql-%Y-%m-%d_%H%M%S.log'
>> ([2]), so I stay that window there, switch to my ssh shell terminal,
>> change the parameter on postgresql.conf, reload pg, come back to
>> pgadmin and ouch! it is frozen!([3]). The same occurs without touch
>> the conf, just waiting with that error window opened... ":(
>>
>
> I thought I was the only one who has this message. I'll work on this.
>

I didn't have this exact message. I can't reproduce your error.

Thanks.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Wed, Feb 25, 2009 at 7:35 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> Fixed with your "truly horrible kludge". The wxTimer is set to half a
> second, and it doesn't enough on Mac. Maybe 400ms would be better.

Doesn't seem too bad to me - my guess was originally for 250ms though.

> Done. I finally understood what's in the perspective. I finally found
> editing the string manually was a better way.

Yeah, I often do that.

> I still didn't change the listview with a textbox for the Logfile pane.
> It's a big work and I would prefer to finish this one before doing the
> change.
>
> See the attached patch.

Now we're cookin' :-). This is much more usable. There are still some
minor issues - let's get these sorted and get it into SVN. We can fix
anything else that crops up from there. Today's testing was performed
on OS X.

- Do we still need all the OnSel... event handlers?

- The log controls (combo box/button) only get activated if a log line
is selected. The should be active whenever the pane is active (if the
log is empty - you can't choose another for example!).

- The cancel query button remained active even when I had selected the log pane.

- The default layout still needs a small tweak. The attached
screenshot is what I get if I remove all frmStatus settings (except
the timers) from the prefs file.

There is also still something weird going on with the Default View
option. Once it's in SVN I'll check that some more as it may be
related to the fact that we don't have the SVN revision in the right
places in frmStatus.h yet.

Tell me - why do you always pick the difficult jobs to work on? :-p

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Wed, Feb 25, 2009 at 7:35 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> Fixed with your "truly horrible kludge". The wxTimer is set to half a
>> second, and it doesn't enough on Mac. Maybe 400ms would be better.
>
> Doesn't seem too bad to me - my guess was originally for 250ms though.
>

I put 400ms, but we can have a lower value if you want.

>> Done. I finally understood what's in the perspective. I finally found
>> editing the string manually was a better way.
>
> Yeah, I often do that.
>
>> I still didn't change the listview with a textbox for the Logfile pane.
>> It's a big work and I would prefer to finish this one before doing the
>> change.
>>
>> See the attached patch.
>
> Now we're cookin' :-). This is much more usable. There are still some
> minor issues - let's get these sorted and get it into SVN. We can fix
> anything else that crops up from there. Today's testing was performed
> on OS X.
>
> - Do we still need all the OnSel... event handlers?
>

Yes, we need them on GTK. On GTK, if I click on a listview, the
associated pane is not activated. I need to click on the title bar of
the pane to really activate the pane.

> - The log controls (combo box/button) only get activated if a log line
> is selected. The should be active whenever the pane is active (if the
> log is empty - you can't choose another for example!).
>

Fixed.

> - The cancel query button remained active even when I had selected the log pane.
>

Fixed.

> - The default layout still needs a small tweak. The attached
> screenshot is what I get if I remove all frmStatus settings (except
> the timers) from the prefs file.
>

I don't have this issue on my MacMini.

> There is also still something weird going on with the Default View
> option. Once it's in SVN I'll check that some more as it may be
> related to the fact that we don't have the SVN revision in the right
> places in frmStatus.h yet.
>

Seems to work on GTK. I'll take a look on my Mac.

> Tell me - why do you always pick the difficult jobs to work on? :-p
>

Because they tend to be the interesting ones? :)

Moreover, if I really wanted to work on hard items, I'll try to work on
the grid view (adding sort and filter on them, something like sqlmanager
can do - see
http://sqlmanager.net/products/postgresql/manager/screenshots/1561).

New patch attached.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Feb 26, 2009 at 12:40 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> I put 400ms, but we can have a lower value if you want.

Meh. Let's set it to 250, but add a micro-optimisation so we don't
reset anything unless the active pane has actually changed.

> Yes, we need them on GTK. On GTK, if I click on a listview, the
> associated pane is not activated. I need to click on the title bar of
> the pane to really activate the pane.

OK.

>> - The default layout still needs a small tweak. The attached
>> screenshot is what I get if I remove all frmStatus settings (except
>> the timers) from the prefs file.
>>
>
> I don't have this issue on my MacMini.

OK.

>> There is also still something weird going on with the Default View
>> option. Once it's in SVN I'll check that some more as it may be
>> related to the fact that we don't have the SVN revision in the right
>> places in frmStatus.h yet.
>>
>
> Seems to work on GTK. I'll take a look on my Mac.

Let's check it in first. It wasn't obvious what was happening, which
is why I didn't look or describe further.

>> Tell me - why do you always pick the difficult jobs to work on? :-p
>>
>
> Because they tend to be the interesting ones? :)

What, like re-writing all the XRC files? :-p

> Moreover, if I really wanted to work on hard items, I'll try to work on
> the grid view (adding sort and filter on them, something like sqlmanager
> can do - see
> http://sqlmanager.net/products/postgresql/manager/screenshots/1561).

You realise we have that already, just not presented in quite the same
way? Try right-clicking a field.

> New patch attached.

Let's get it in to SVN now. The code is pretty solid now, and we can
fix any bugs one by one. Plus, our QA team are testing daily at the
moment so they'll be able to spot bugs as well.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Thu, Feb 26, 2009 at 12:40 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> I put 400ms, but we can have a lower value if you want.
>
> Meh. Let's set it to 250, but add a micro-optimisation so we don't
> reset anything unless the active pane has actually changed.
>

Done.

>> Yes, we need them on GTK. On GTK, if I click on a listview, the
>> associated pane is not activated. I need to click on the title bar of
>> the pane to really activate the pane.
>
> OK.
>
>>> - The default layout still needs a small tweak. The attached
>>> screenshot is what I get if I remove all frmStatus settings (except
>>> the timers) from the prefs file.
>>>
>> I don't have this issue on my MacMini.
>
> OK.
>
>>> There is also still something weird going on with the Default View
>>> option. Once it's in SVN I'll check that some more as it may be
>>> related to the fact that we don't have the SVN revision in the right
>>> places in frmStatus.h yet.
>>>
>> Seems to work on GTK. I'll take a look on my Mac.
>
> Let's check it in first. It wasn't obvious what was happening, which
> is why I didn't look or describe further.
>
>>> Tell me - why do you always pick the difficult jobs to work on? :-p
>>>
>> Because they tend to be the interesting ones? :)
>
> What, like re-writing all the XRC files? :-p
>

Well, rewriting XRC files is no fun at all :) but the results are more
then welcome (at least to me).

>> Moreover, if I really wanted to work on hard items, I'll try to work on
>> the grid view (adding sort and filter on them, something like sqlmanager
>> can do - see
>> http://sqlmanager.net/products/postgresql/manager/screenshots/1561).
>
> You realise we have that already, just not presented in quite the same
> way? Try right-clicking a field.
>

Doesn't work on the configuration editor which really is *the* UI I'm
completely unable to use.

>> New patch attached.
>
> Let's get it in to SVN now. The code is pretty solid now, and we can
> fix any bugs one by one. Plus, our QA team are testing daily at the
> moment so they'll be able to spot bugs as well.
>

Applied, thanks.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Guillaume Lelarge a écrit :
> Dave Page a écrit :
>> Forgot to mention - what happened about changing the log pane to use a
>> text control?
>>
>
> It seems I forgot it. Sorry.
>
>

I took a look at this. It doesn't seem to me that it will requires a lot
of work. But using a text control will kill an interesting functionality
(which needs more love, but still the functionality is interesting).

If you set the log_line_prefix to a specific format, the listview gets
two more columns (timestamp and level). If we change the listview
control with a text control, we will loose this functionality. I would
much prefer to keep it.

Why do you want to have a text control? to copy part of the log text?

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Feb 26, 2009 at 4:43 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:

> I took a look at this. It doesn't seem to me that it will requires a lot
> of work. But using a text control will kill an interesting functionality
> (which needs more love, but still the functionality is interesting).
>
> If you set the log_line_prefix to a specific format, the listview gets
> two more columns (timestamp and level). If we change the listview
> control with a text control, we will loose this functionality. I would
> much prefer to keep it.
>
> Why do you want to have a text control? to copy part of the log text?

Yeah - that was the idea (there's also a bug on Windows - each line
ends with a square, which I assume is the <CR>).

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Thu, Feb 26, 2009 at 4:43 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>
>> I took a look at this. It doesn't seem to me that it will requires a lot
>> of work. But using a text control will kill an interesting functionality
>> (which needs more love, but still the functionality is interesting).
>>
>> If you set the log_line_prefix to a specific format, the listview gets
>> two more columns (timestamp and level). If we change the listview
>> control with a text control, we will loose this functionality. I would
>> much prefer to keep it.
>>
>> Why do you want to have a text control? to copy part of the log text?
>
> Yeah - that was the idea (there's also a bug on Windows - each line
> ends with a square, which I assume is the <CR>).
>

Do you need to copy multiple lines? we can use the copy item of the edit
menu to allow one to copy one or more lines? if you're OK, I'll make a
patch for this and the Windows bug.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

On Thu, Feb 26, 2009 at 4:49 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Dave Page a écrit :
>> On Thu, Feb 26, 2009 at 4:43 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>>
>>> I took a look at this. It doesn't seem to me that it will requires a lot
>>> of work. But using a text control will kill an interesting functionality
>>> (which needs more love, but still the functionality is interesting).
>>>
>>> If you set the log_line_prefix to a specific format, the listview gets
>>> two more columns (timestamp and level). If we change the listview
>>> control with a text control, we will loose this functionality. I would
>>> much prefer to keep it.
>>>
>>> Why do you want to have a text control? to copy part of the log text?
>>
>> Yeah - that was the idea (there's also a bug on Windows - each line
>> ends with a square, which I assume is the <CR>).
>>
>
> Do you need to copy multiple lines? we can use the copy item of the edit
> menu to allow one to copy one or more lines? if you're OK, I'll make a
> patch for this and the Windows bug.

Sure, that would work. It'd be nice to be able to copy any of the
listviews in fact (wouldn't be hard to do I guess - just iterate the
rows and build a string with tab delimited columns and <lf> delimited
rows).

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

OS: Ubuntu 8.04 Linux analise3 2.6.25.15-bd-mod #1 SMP PREEMPT Thu Nov
27 10:05:44 BRST 2008 i686
SVN Revision: 7614

Problem 1: Whithout adminpack.sql loaded to db, pgadmin segfault when
open Status Window:

Backtrace:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb6471990 (LWP 9723)]
0xb77f1df5 in wxGenericListCtrl::GetItemCount () from
/usr/lib/libwx_gtk2u_core-2.8.so.0
(gdb) bt
#0 0xb77f1df5 in wxGenericListCtrl::GetItemCount () from
/usr/lib/libwx_gtk2u_core-2.8.so.0
#1 0x080b6253 in ctlListView::AppendItem ()
#2 0x082a8183 in frmStatus::OnRefreshLogTimer ()
#3 0x082a8a3d in frmStatus::OnRefresh ()
#4 0x082ab7af in frmStatus::OnRateChange ()
#5 0x082ab9da in frmStatus::Go ()
#6 0x082b7e00 in serverStatusFactory::StartDialog ()
#7 0x081db886 in frmMain::OnAction ()
#8 0xb75360e1 in wxAppConsole::HandleEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#9 0xb75ddb88 in wxEvtHandler::ProcessEventIfMatches () from
/usr/lib/libwx_baseu-2.8.so.0
#10 0xb75ddd74 in wxEvtHandler::SearchDynamicEventTable () from
/usr/lib/libwx_baseu-2.8.so.0
#11 0xb75dde33 in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#12 0xb75dddfb in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#13 0xb7843e73 in ?? () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#14 0xb6c06aef in g_cclosure_marshal_VOID__VOID () from
/usr/lib/libgobject-2.0.so.0
#15 0xb6bf9749 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#16 0xb6c0df7b in ?? () from /usr/lib/libgobject-2.0.so.0
#17 0xb6c0fc0f in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#18 0xb6c0ff59 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#19 0xb6f8f278 in gtk_widget_activate () from /usr/lib/libgtk-x11-2.0.so.0
#20 0xb6e78bc2 in gtk_menu_shell_activate_item () from
/usr/lib/libgtk-x11-2.0.so.0
#21 0xb6e7a708 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#22 0xb6e71914 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#23 0xb6e6b8d4 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#24 0xb6bf8069 in ?? () from /usr/lib/libgobject-2.0.so.0
#25 0xb6bf9749 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#26 0xb6c0e10a in ?? () from /usr/lib/libgobject-2.0.so.0
#27 0xb6c0f937 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#28 0xb6c0ff59 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#29 0xb6f8a667 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#30 0xb6e64b21 in gtk_propagate_event () from /usr/lib/libgtk-x11-2.0.so.0
#31 0xb6e65d88 in gtk_main_do_event () from /usr/lib/libgtk-x11-2.0.so.0
#32 0xb6cdda9a in ?? () from /usr/lib/libgdk-x11-2.0.so.0
#33 0xb6b62cc6 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#34 0xb6b66083 in ?? () from /usr/lib/libglib-2.0.so.0
#35 0xb6b66467 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#36 0xb6e66264 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#37 0xb77c071c in wxEventLoop::Run () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#38 0xb786332e in wxAppBase::MainLoop () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#39 0xb7862981 in wxAppBase::OnRun () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#40 0xb757205a in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#41 0xb7572107 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#42 0x0807ca70 in main ()

[]s
Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

Hi all,

OS: Ubuntu 8.04 Linux analise3 2.6.25.15-bd-mod #1 SMP PREEMPT Thu Nov
27 10:05:44 BRST 2008 i686
SVN Revision: 7614

Problem 2: With adminpack.sql loaded, when "logging_colector not
enabled or log_filename misconfigured" is the only item in the Logfile
ListBox (because I don't configured it yet), I click it then ComboBox
in right of "Rotate" button is Enabled, I choose "current" and a
segfault occurs.

Log:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb6475990 (LWP 9991)]
0x0846d998 in DateToAnsiStr ()
Current language: auto; currently asm
(gdb) bt
#0 0x0846d998 in DateToAnsiStr ()
#1 0x082a7b25 in frmStatus::addLogFile ()
#2 0x082a7f90 in frmStatus::OnLoadLogfile ()
#3 0xb753a0e1 in wxAppConsole::HandleEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#4 0xb75e1b88 in wxEvtHandler::ProcessEventIfMatches () from
/usr/lib/libwx_baseu-2.8.so.0
#5 0xb75e1ce8 in wxEventHashTable::HandleEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#6 0xb75e1e4f in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#7 0xb75e1dfb in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#8 0xb78f996e in wxWindowBase::TryParent () from
/usr/lib/libwx_gtk2u_core-2.8.so.0
#9 0xb75e1e0b in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#10 0xb78f996e in wxWindowBase::TryParent () from
/usr/lib/libwx_gtk2u_core-2.8.so.0
#11 0xb75e1e0b in wxEvtHandler::ProcessEvent () from
/usr/lib/libwx_baseu-2.8.so.0
#12 0xb7826894 in ?? () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#13 0xb6c0aaef in g_cclosure_marshal_VOID__VOID () from
/usr/lib/libgobject-2.0.so.0
#14 0xb6bfd749 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#15 0xb6c12319 in ?? () from /usr/lib/libgobject-2.0.so.0
#16 0xb6c13c0f in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#17 0xb6c13f59 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#18 0xb6dc9901 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#19 0xb6dcadb2 in gtk_combo_box_set_active_iter () from
/usr/lib/libgtk-x11-2.0.so.0
#20 0xb6dcaece in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#21 0xb6c0aaef in g_cclosure_marshal_VOID__VOID () from
/usr/lib/libgobject-2.0.so.0
#22 0xb6bfd749 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#23 0xb6c11f7b in ?? () from /usr/lib/libgobject-2.0.so.0
#24 0xb6c13c0f in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#25 0xb6c13f59 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#26 0xb6f93278 in gtk_widget_activate () from /usr/lib/libgtk-x11-2.0.so.0
#27 0xb6e7cbc2 in gtk_menu_shell_activate_item () from
/usr/lib/libgtk-x11-2.0.so.0
#28 0xb6e7e708 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#29 0xb6e75914 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#30 0xb6e6f8d4 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#31 0xb6bfc069 in ?? () from /usr/lib/libgobject-2.0.so.0
#32 0xb6bfd749 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#33 0xb6c1210a in ?? () from /usr/lib/libgobject-2.0.so.0
#34 0xb6c13937 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0
#35 0xb6c13f59 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#36 0xb6f8e667 in ?? () from /usr/lib/libgtk-x11-2.0.so.0
#37 0xb6e68b21 in gtk_propagate_event () from /usr/lib/libgtk-x11-2.0.so.0
#38 0xb6e69d88 in gtk_main_do_event () from /usr/lib/libgtk-x11-2.0.so.0
#39 0xb6ce1a9a in ?? () from /usr/lib/libgdk-x11-2.0.so.0
#40 0xb6b66cc6 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#41 0xb6b6a083 in ?? () from /usr/lib/libglib-2.0.so.0
#42 0xb6b6a467 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
#43 0xb6e6a264 in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#44 0xb77c471c in wxEventLoop::Run () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#45 0xb786732e in wxAppBase::MainLoop () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#46 0xb7866981 in wxAppBase::OnRun () from /usr/lib/libwx_gtk2u_core-2.8.so.0
#47 0xb757605a in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#48 0xb7576107 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#49 0x0807ca70 in main ()

I hope this help...

Regards,

Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

Dave Page a écrit :
> On Thu, Feb 26, 2009 at 4:49 PM, Guillaume Lelarge
> <guillaume(at)lelarge(dot)info> wrote:
>> Dave Page a écrit :
>>> On Thu, Feb 26, 2009 at 4:43 PM, Guillaume Lelarge
>>> <guillaume(at)lelarge(dot)info> wrote:
>>>
>>>> I took a look at this. It doesn't seem to me that it will requires a lot
>>>> of work. But using a text control will kill an interesting functionality
>>>> (which needs more love, but still the functionality is interesting).
>>>>
>>>> If you set the log_line_prefix to a specific format, the listview gets
>>>> two more columns (timestamp and level). If we change the listview
>>>> control with a text control, we will loose this functionality. I would
>>>> much prefer to keep it.
>>>>
>>>> Why do you want to have a text control? to copy part of the log text?
>>> Yeah - that was the idea (there's also a bug on Windows - each line
>>> ends with a square, which I assume is the <CR>).
>>>
>> Do you need to copy multiple lines? we can use the copy item of the edit
>> menu to allow one to copy one or more lines? if you're OK, I'll make a
>> patch for this and the Windows bug.
>
> Sure, that would work. It'd be nice to be able to copy any of the
> listviews in fact (wouldn't be hard to do I guess - just iterate the
> rows and build a string with tab delimited columns and <lf> delimited
> rows).
>

This patch does this : it copies every columns of every selected rows of
the active pane.

I didn't work on the square issue because I have a problem to compile
under Windows. I think it's due to the frmStatus.xrc file deletion but
can't find how to fix this. I'll look more tomorrow.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

Dickson S. Guedes a écrit :
> OS: Ubuntu 8.04 Linux analise3 2.6.25.15-bd-mod #1 SMP PREEMPT Thu Nov
> 27 10:05:44 BRST 2008 i686
> SVN Revision: 7614
>
> Problem 1: Whithout adminpack.sql loaded to db, pgadmin segfault when
> open Status Window:
>

It seems you're right. It doesn't segfault with me, but it has a bad
behavior. I'll work on this (and your other mail) tomorrow.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

2009/2/26 Dickson S. Guedes <listas(at)guedesoft(dot)net>:
> OS: Ubuntu 8.04 Linux analise3 2.6.25.15-bd-mod #1 SMP PREEMPT Thu Nov
> 27 10:05:44 BRST 2008 i686
> SVN Revision: 7614
>
> Problem 2: With adminpack.sql loaded, when "logging_colector not
> enabled or log_filename misconfigured" is the only item in the Logfile
> ListBox (because I don't configured it yet), I click it then ComboBox
> in right of "Rotate" button is Enabled, I choose "current" and a
> segfault occurs.

Hi Guillaume,

May be the attached patch could be useful for this problem. I hope it helps.

I'm using g++ (GCC) 4.2.4 (Ubuntu 4.2.4-1ubuntu3).

Regards.

Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

On Thu, Feb 26, 2009 at 12:40 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Dave Page a écrit :
>> - The default layout still needs a small tweak. The attached
>> screenshot is what I get if I remove all frmStatus settings (except
>> the timers) from the prefs file.
>>
>
> I don't have this issue on my MacMini.

FYI, the default on Windows looks pretty similar to the Mac screenshot
I sent. See attached (give me a few minutes before testing - I'm just
about to commit a tweaked version of the copy patch).

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

On Thu, Feb 26, 2009 at 6:45 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Dave Page a écrit :

>> Sure, that would work. It'd be nice to be able to copy any of the
>> listviews in fact (wouldn't be hard to do I guess - just iterate the
>> rows and build a string with tab delimited columns and <lf> delimited
>> rows).
>>
>
> This patch does this : it copies every columns of every selected rows of
> the active pane.

Thanks - patch applied, with minor tweaks to remove the unused edit
menu options (let's not clutter menus with completely unused options),
and to use \r\n as the line end on Windows (so we can paste into
Notepad).

> I didn't work on the square issue because I have a problem to compile
> under Windows. I think it's due to the frmStatus.xrc file deletion but
> can't find how to fix this. I'll look more tomorrow.

I've fixed the square issue by Trim()ing log lines before adding them.

Thanks!

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dickson S. Guedes a écrit :
> 2009/2/26 Dickson S. Guedes <listas(at)guedesoft(dot)net>:
>> OS: Ubuntu 8.04 Linux analise3 2.6.25.15-bd-mod #1 SMP PREEMPT Thu Nov
>> 27 10:05:44 BRST 2008 i686
>> SVN Revision: 7614
>>
>> Problem 2: With adminpack.sql loaded, when "logging_colector not
>> enabled or log_filename misconfigured" is the only item in the Logfile
>> ListBox (because I don't configured it yet), I click it then ComboBox
>> in right of "Rotate" button is Enabled, I choose "current" and a
>> segfault occurs.
>
> Hi Guillaume,
>
> May be the attached patch could be useful for this problem. I hope it helps.
>

It helps a lot. It fixed problem 2. Thanks for your report and patch.

Your problem #1 should be fixed too with my latest commit.

Thanks a lot.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com

2009/2/27 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
> It helps a lot. It fixed problem 2. Thanks for your report and patch.

Nice! ":D

> Your problem #1 should be fixed too with my latest commit.

I've tried now with SVN revision 7621, and AFAICS they are working ok, thanks!

Now when I select itens, in Locks UI for example, then "Copy" using
the button, the selected list changes because a refresh occurs between
I select and click, so, the final copied text isn't the same that I
orinally choose.

":(

[]s
Dickson S. Guedes
-
mail/xmpp: guedes(at)guedesoft(dot)net - skype: guediz
http://guedesoft.net - http://planeta.postgresql.org.br

Dickson S. Guedes a écrit :
> 2009/2/27 Guillaume Lelarge <guillaume(at)lelarge(dot)info>:
>> It helps a lot. It fixed problem 2. Thanks for your report and patch.
>
> Nice! ":D
>
>> Your problem #1 should be fixed too with my latest commit.
>
> I've tried now with SVN revision 7621, and AFAICS they are working ok, thanks!
>
> Now when I select itens, in Locks UI for example, then "Copy" using
> the button, the selected list changes because a refresh occurs between
> I select and click, so, the final copied text isn't the same that I
> orinally choose.
>

I don't think there's something we can do about this. We can't stop
refreshing when a line is selected.

Ideas welcome :)

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com