Re: [HACKERS] commitfest.postgresql.org

Per Dave Page's request of this morning, my CommitFest management
application now has a real hostname (see subject line). I have also
sent Dave an email with details of the install process and location of
files, per his request (let me know if there's somewhere else those
details should be posted).

Brendan Jurd has graciously migrated all of the data from the
CommitFest wiki page to the app by writing a script to parse the wiki
markup and inserting the resulting data directly into the database.
There are a few loose ends. The application stamps comments with the
community login of the person who left them, but the import stamped
them with names instead. This is actually of some significance, since
the app will allow you to edit your own comments but not those of
other people. We could probably fix this if someone can give us
access to (or a dump of) the realname to username mappings from the
community login DB. Also, we're currently missing the reviewer names
due to limitations of the import script; Brendan is fixing this.

Filling the DB with live data revealed a few warts. In particular,
the original ordering of patches was alpha by topic and then alpha by
name, which I thought would be OK, but upon seeing how it really
looked, I hated it. So the topic manager now lets you set a sort
order for commitfest topics, and the order is now numeric by topic
sortorder, then alpha by topic, then by ascending patch ID (so the
oldest patch comes up first within each topic section). Also, I
originally had the topics displayed as a table column, but that didn't
really work for me once I saw it either, so it's been reorganized to
do it the same way the wiki does.

We're still hacking on a few other details of the formatting and
interface, but you might want to cruise over and have a look. Please
however continue to make ALL CHANGES on the wiki, and not in the app.
Brendan is going to manually move changes over to the new system
incrementally until we get the kinks worked out. I think we're close,
but we're not quite there yet. As a reminder, if you'd like to review
or submit patches against the source, you can find it here:

http://git.postgresql.org/gitweb?p=pgcommitfest.git;a=summary
git://git.postgresql.org/git/pgcommitfest.git

...Robert

2009/7/3 Robert Haas <robertmhaas(at)gmail(dot)com>:
> Also, we're currently missing the reviewer names
> due to limitations of the import script; Brendan is fixing this.

Update: The reviewer names have now been populated.

Cheers,
BJ

On Fri, Jul 3, 2009 at 12:34 AM, Brendan Jurd<direvus(at)gmail(dot)com> wrote:
> 2009/7/3 Robert Haas <robertmhaas(at)gmail(dot)com>:
>> Also, we're currently missing the reviewer names
>> due to limitations of the import script; Brendan is fixing this.
>
> Update: The reviewer names have now been populated.
>

looks good... now, how can i mark a patch as being reviewed? or i have
to told you in order to you mark it?

--
Atentamente,
Jaime Casanova
Soporte y capacitación de PostgreSQL
Asesoría y desarrollo de sistemas
Guayaquil - Ecuador
Cel. +59387171157

2009/7/3 Jaime Casanova <jcasanov(at)systemguards(dot)com(dot)ec>:
> On Fri, Jul 3, 2009 at 12:34 AM, Brendan Jurd<direvus(at)gmail(dot)com> wrote:
>> 2009/7/3 Robert Haas <robertmhaas(at)gmail(dot)com>:
>>> Also, we're currently missing the reviewer names
>>> due to limitations of the import script; Brendan is fixing this.
>>
>> Update: The reviewer names have now been populated.
>>
>
> looks good... now, how can i mark a patch as being reviewed? or i have
> to told you in order to you mark it?

Until we have a solid consensus to switch to using the new app, please
continue to make your changes on the wiki. I'll replicate the changes
by hand as we go until switchover.

If you'd like to make your changes on the app yourself and save me the
trouble, please do! Just select the patch you're interested in, click
"edit patch" on the upper right and change the status and/or reviewer
fields as required.

Please let us know if you encounter any problems with withe app, or
have suggestions for improvement.

Cheers,
BJ

Brendan Jurd <direvus(at)gmail(dot)com> writes:
> Please let us know if you encounter any problems with withe app, or
> have suggestions for improvement.

It looks like every patch and comment is timestamped ... but with
yesterday (the time of data import, I suppose). This is much worse
than useless. If you can't get the actual date, leave it off.

regards, tom lane

On "suggestions for improvement": I need to be able to bookmark the
commitfest summary list (whichever page is equivalent to the old wiki
page). The current URL seems to be
http://commitfest.postgresql.org/action/commitfest_view?id=2
which is both opaque as can be and not looking like it's intended to
be stable over the long term. I can also imagine people wanting to
refer to particular patch entries in email, but those URLs are no
better. Could we pay some attention to using URLs that are stable
and self-explanatory?

regards, tom lane

I tried the "New Patch Comment" feature. It's absolutely horrid.
I get a page showing a "comment type" button, one line for Message-ID,
and one line for Content. No explanation of what those are, and no
visibility any more of the patch I'm trying to comment on. I have no
idea what I'm supposed to do here, and if I were trying to respond to
someone else's comment, it would be nice to be able to see that comment.

regards, tom lane

On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> On "suggestions for improvement": I need to be able to bookmark the
> commitfest summary list (whichever page is equivalent to the old wiki
> page).  The current URL seems to be
> http://commitfest.postgresql.org/action/commitfest_view?id=2
> which is both opaque as can be and not looking like it's intended to
> be stable over the long term.  I can also imagine people wanting to
> refer to particular patch entries in email, but those URLs are no
> better.  Could we pay some attention to using URLs that are stable
> and self-explanatory?

I'm not sure why you would think that it's not stable. It's not
stable in the sense that it won't always be the open CommitFest, but
it's certainly stable in the sense that it will always be the same
CommitFest that it is now. It's actually MORE stable than the wiki,
since the wiki doesn't permit you to change the name of the CommitFest
without also changing the URL.

I'm also not sure what you would think that it's not self-explanatory,
since it looks pretty self explanatory to me. You've asked to view a
commitfest. The id of that commitfest is 2. What more do you want to
know?

If you're advocating for the use of wiki-style names, where the URL
actually contains the name of the things that it points to, then you
have incompatible requirements, because things can, do, and will
continue to get renamed. If the URL is built around the name, then
the URL will change when the name changes. If you want it to be
stable, a non-natural key is your only option. And I frankly don't
see what's wrong with that. We regularly deal in message-IDs or other
links to the archives, which are certainly totally opaque.
Fortunately, they're links. You can click on them, and then you see
what they are.

...Robert

On Fri, Jul 3, 2009 at 10:40 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> I tried the "New Patch Comment" feature.  It's absolutely horrid.
> I get a page showing a "comment type" button, one line for Message-ID,
> and one line for Content.  No explanation of what those are, and no
> visibility any more of the patch I'm trying to comment on.  I have no
> idea what I'm supposed to do here, and if I were trying to respond to
> someone else's comment, it would be nice to be able to see that comment.

The message-ID is the (optional) ID of a message you'd like to link
to. The comment is the text of your comment. If there's a legitimate
reason for confusion here, I have no idea what it is.

I agree that the comment box probably needs to be converted into a
text area rather than a single line. But I also think that comments
on the wiki should be kept short. If you have more than a few lines
of text, there's a good chance you should be sending an email to
-hackers and then linking to it. There are some projects that are
managed using a discussion forum or a bug tracker and as far as I know
you've always been opposed to that, as am I. So complaining that this
system doesn't work that way seems disingenuous. Also, if you think
this interface is inconvenient for leaving a comment, have you tried
doing it on the wiki lately?

Anyway, I'm sure there's room for improvement in this tool. I intend
to make improvements. I still think it's already better than the
wiki. Moving things to a different commitfest or a different section
of the same commitfest (like, particularly relevant to you, the
"committed" section) is takes me minutes on the wiki and seconds with
this app.

...Robert

Robert Haas wrote:
> On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> On "suggestions for improvement": I need to be able to bookmark the
>> commitfest summary list (whichever page is equivalent to the old wiki
>> page). The current URL seems to be
>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>> which is both opaque as can be and not looking like it's intended to
>> be stable over the long term. I can also imagine people wanting to
>> refer to particular patch entries in email, but those URLs are no
>> better. Could we pay some attention to using URLs that are stable
>> and self-explanatory?
>
> I'm not sure why you would think that it's not stable. It's not
> stable in the sense that it won't always be the open CommitFest, but
> it's certainly stable in the sense that it will always be the same
> CommitFest that it is now. It's actually MORE stable than the wiki,
> since the wiki doesn't permit you to change the name of the CommitFest
> without also changing the URL.

We have redirects to the previous, open, and in-progress commitfests in
the wiki (see http://wiki.postgresql.org/wiki/CommitFest). Those
redirects are bookmarkable.

--
Heikki Linnakangas
EnterpriseDB http://www.enterprisedb.com

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> The current URL seems to be
>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>> which is both opaque as can be and not looking like it's intended to
>> be stable over the long term.

> I'm not sure why you would think that it's not stable.

Because it's exposing three or four details of your implementation,
which you might wish to change later.

> I'm also not sure what you would think that it's not self-explanatory,
> since it looks pretty self explanatory to me.

It's impossible to know that this is commitfest 2009-07.

regards, tom lane

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Fri, Jul 3, 2009 at 10:40 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> I tried the "New Patch Comment" feature. It's absolutely horrid.
>> I get a page showing a "comment type" button, one line for Message-ID,
>> and one line for Content. No explanation of what those are, and no
>> visibility any more of the patch I'm trying to comment on. I have no
>> idea what I'm supposed to do here, and if I were trying to respond to
>> someone else's comment, it would be nice to be able to see that comment.

> The message-ID is the (optional) ID of a message you'd like to link
> to. The comment is the text of your comment. If there's a legitimate
> reason for confusion here, I have no idea what it is.

It's not apparent that the message-ID is optional, nor what it is for
really. And the fact that the boxes are the same size leaves one
wondering what the comment is supposed to be too. The basic complaint
is that the page assumes you already know what to do with it.
Given the vast amount of white space left behind by the omission of
any context information, it doesn't seem unreasonable to include a
couple sentences of explanation.

> I agree that the comment box probably needs to be converted into a
> text area rather than a single line. But I also think that comments
> on the wiki should be kept short. If you have more than a few lines
> of text, there's a good chance you should be sending an email to
> -hackers and then linking to it. There are some projects that are
> managed using a discussion forum or a bug tracker and as far as I know
> you've always been opposed to that, as am I. So complaining that this
> system doesn't work that way seems disingenuous. Also, if you think
> this interface is inconvenient for leaving a comment, have you tried
> doing it on the wiki lately?

I spent a large part of the last year leaving comments on the wiki.
Yeah, it was a bit tedious to use wiki markup, but at least all the
information you needed was a click away. (The wiki wasn't designed
on the assumption that users already know how to use it.) And the
context didn't all disappear from view the moment you wanted to add
something.

> I still think it's already better than the wiki.

Maybe to you, but right now I think the wiki is far more usable
and far more flexible. It hasn't got arbitrary assumptions about
what size comment people are allowed to leave, for example.

regards, tom lane

On Fri, Jul 3, 2009 at 11:46 AM, Heikki
Linnakangas<heikki(dot)linnakangas(at)enterprisedb(dot)com> wrote:
> Robert Haas wrote:
>> On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> On "suggestions for improvement": I need to be able to bookmark the
>>> commitfest summary list (whichever page is equivalent to the old wiki
>>> page).  The current URL seems to be
>>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>>> which is both opaque as can be and not looking like it's intended to
>>> be stable over the long term.  I can also imagine people wanting to
>>> refer to particular patch entries in email, but those URLs are no
>>> better.  Could we pay some attention to using URLs that are stable
>>> and self-explanatory?
>>
>> I'm not sure why you would think that it's not stable.  It's not
>> stable in the sense that it won't always be the open CommitFest, but
>> it's certainly stable in the sense that it will always be the same
>> CommitFest that it is now.  It's actually MORE stable than the wiki,
>> since the wiki doesn't permit you to change the name of the CommitFest
>> without also changing the URL.
>
> We have redirects to the previous, open, and in-progress commitfests in
> the wiki (see http://wiki.postgresql.org/wiki/CommitFest). Those
> redirects are bookmarkable.

We can do the same thing here. It's a SMOP.

...Robert

On Friday 03 July 2009 11:50:29 Tom Lane wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >> The current URL seems to be
> >> http://commitfest.postgresql.org/action/commitfest_view?id=2
> >> which is both opaque as can be and not looking like it's intended to
> >> be stable over the long term.
> >
> > I'm not sure why you would think that it's not stable.
>
> Because it's exposing three or four details of your implementation,
> which you might wish to change later.
>
> > I'm also not sure what you would think that it's not self-explanatory,
> > since it looks pretty self explanatory to me.
>
> It's impossible to know that this is commitfest 2009-07.
>

commitfest.postgresql.org/2009/07 ?

That, or any similar scheme, seems easily doable with a little apache rewrite
magic and some programming. See my blog urls for one such example.

--
Robert Treat
Conjecture: http://www.xzilla.net
Consulting: http://www.omniti.com

On Fri, 2009-07-03 at 14:06 -0400, Robert Treat wrote:
> On Friday 03 July 2009 11:50:29 Tom Lane wrote:
> > Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > > On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > >> The current URL seems to be
> > >> http://commitfest.postgresql.org/action/commitfest_view?id=2
> > >> which is both opaque as can be and not looking like it's intended to
> > >> be stable over the long term.
> > >
> > > I'm not sure why you would think that it's not stable.
> >
> > Because it's exposing three or four details of your implementation,
> > which you might wish to change later.
> >
> > > I'm also not sure what you would think that it's not self-explanatory,
> > > since it looks pretty self explanatory to me.
> >
> > It's impossible to know that this is commitfest 2009-07.
> >
>
> commitfest.postgresql.org/2009/07 ?
>
> That, or any similar scheme, seems easily doable with a little apache rewrite
> magic and some programming. See my blog urls for one such example.

O.k. I am probably blowing something out of the water here but do we
need yet another domain?

wiki.postgresql.org/commitfest/2009/07 (or www)

Joshua D. Drake

--
PostgreSQL - XMPP: jdrake(at)jabber(dot)postgresql(dot)org
Consulting, Development, Support, Training
503-667-4564 - http://www.commandprompt.com/
The PostgreSQL Company, serving since 1997

On Fri, Jul 3, 2009 at 11:50 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> The current URL seems to be
>>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>>> which is both opaque as can be and not looking like it's intended to
>>> be stable over the long term.
>
>> I'm not sure why you would think that it's not stable.
>
> Because it's exposing three or four details of your implementation,
> which you might wish to change later.
>
>> I'm also not sure what you would think that it's not self-explanatory,
>> since it looks pretty self explanatory to me.
>
> It's impossible to know that this is commitfest 2009-07.

Backing up for a moment to ten thousand feet here, I posted a link to
this web app on May 26th. I received several comments on it, all of
them positive, including some constructive feedback from you which I
took to heart. It is now July 1st, and I am trying very hard to get
ready for the next CommitFest, which I have agreed to manage. So I
need to determine whether there is some finite number of changes of
manageable size that I can make to get this to a state where we can
use it, or whether I should give up hope now and go back to the wiki.
The latter outcome would be disappointing to me, but that's where
we're going to end up if we can't actually boil this down to a
specific list of changes to be made.

I accept the need for and am willing to make the following changes:

- Changing the patch comment field from type text to type textarea and
integrating it into the patch view page to provide context.
- Adding a note to the effect that the message ID is optional.
- Adding stable links with mnemonic names for the open, in progress,
and most recently closed commitfests.

With respect to the issue of the page URLs, I'm very unconvinced of
the value of making a change. First of all, one of your arguments is
that I might want to change them later. I can promise you that I
won't want to do any such thing, or at the VERY least that the old
URLs will always be supported. There would be a sort of perverse
irony to me changing the application to use a different kind of key
for the reason that Tom Lane is afraid that some day I might want to
use a different kind of key.

Secondly, if we are going to make a change, it would be nice to know
what the use case is for the change and to have a set of requirements
that are not mutually contradictory. It is not possible to both
assign the URLs to the pages based on the names of the objects to
which the refer (which are changeable) and to guarantee that the URLs
will never change. So you can either have URL stability or you can
have wiki-style names, but you can't have both, unless perhaps you
include both the ID and the name in the link but make the name just
decoration. I am frankly at a loss to why this is a big deal: if you
bookmark the page, your browser will record the page title; if you use
firefox, you can also find by page title in the address bar. And
we've never had pages for individual patches before anyway, so why
should it now be critical how those links are formatted? But if it is
critical then please describe exactly how you think it should work.

...Robert

On Fri, Jul 3, 2009 at 2:10 PM, Joshua D. Drake<jd(at)commandprompt(dot)com> wrote:
> O.k. I am probably blowing something out of the water here but do we
> need yet another domain?

Because it's installed on a different VM and I don't want to move it
just to make the URL look different?

...Robert

Robert Treat wrote:
> On Friday 03 July 2009 11:50:29 Tom Lane wrote:
>> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>>> On Fri, Jul 3, 2009 at 10:35 AM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>>> The current URL seems to be
>>>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>>>> which is both opaque as can be and not looking like it's intended to
>>>> be stable over the long term.
>>> I'm not sure why you would think that it's not stable.
>> Because it's exposing three or four details of your implementation,
>> which you might wish to change later.
>>
>>> I'm also not sure what you would think that it's not self-explanatory,
>>> since it looks pretty self explanatory to me.
>> It's impossible to know that this is commitfest 2009-07.
>>
>
> commitfest.postgresql.org/2009/07 ?
>
> That, or any similar scheme, seems easily doable with a little apache rewrite
> magic and some programming. See my blog urls for one such example.
>

I believe Tom wants details removed from the URL, so future
implementation changes don't either a) break bookmarks because more
stuff is needed in the URL or b) don't break bookmarks but be limited to
existing sutff in the URL (ie. hacky work arounds). If that's the case,
your best best is to use some kind of key, like 16 random bytes
displayed in hex, that looks up the data.

IMHO, I don't see much gain to encoding the date into the url either.
This is not a great way of telling the user when something occurred. A
lookup is going to occur either way, so why not get all data at once
using a single method?

--
Andrew Chernow
eSilo, LLC
every bit counts
http://www.esilo.com/

On Fri, Jul 3, 2009 at 3:00 PM, Andrew Chernow<ac(at)esilo(dot)com> wrote:
>>>>> The current URL seems to be
>>>>> http://commitfest.postgresql.org/action/commitfest_view?id=2
>>>>> which is both opaque as can be and not looking like it's intended to
>>>>> be stable over the long term.
>>>>
>>>> I'm not sure why you would think that it's not stable.
>>>
>>> Because it's exposing three or four details of your implementation,
>>> which you might wish to change later.
>>>
>>>> I'm also not sure what you would think that it's not self-explanatory,
>>>> since it looks pretty self explanatory to me.
>>>
>>> It's impossible to know that this is commitfest 2009-07.
>>>
>>
>> commitfest.postgresql.org/2009/07 ?
>>
>> That, or any similar scheme, seems easily doable with a little apache
>> rewrite magic and some programming. See my blog urls for one such example.
>
> I believe Tom wants details removed from the URL, so future implementation
> changes don't either a) break bookmarks because more stuff is needed in the
> URL or b) don't break bookmarks but be limited to existing sutff in the URL
> (ie. hacky work arounds).  If that's the case, your best best is to use some
> kind of key, like 16 random bytes displayed in hex, that looks up the data.

I *am* using some kind of key. Specifically, in integer derived from
a serial column. It's just as stable as 16 random bytes displayed in
hex, but a lot shorter and easier to remember, if you're the sort of
person who likes to remember URLs. :-)

> IMHO, I don't see much gain to encoding the date into the url either. This
> is not a great way of telling the user when something occurred.  A lookup is
> going to occur either way, so why not get all data at once using a single
> method?

Sorry, I'm not following this part.

...Robet

Hi,

Le 3 juil. 09 à 20:44, Robert Haas a écrit :
> - Adding stable links with mnemonic names for the open, in progress,
> and most recently closed commitfests.

May I suggest something looking about like:
http://commitfest.postgresql.org/current
http://commitfest.postgresql.org/open
http://commitfest.postgresql.org/in-progress
http://commitfest.postgresql.org/previous

> With respect to the issue of the page URLs, I'm very unconvinced of
> the value of making a change.

Your software seems to be better than a wiki, but its potential users
are expressing needs and bikescheding. I think you'd better accept
both kind of changes as long as it's not making your life much harder
than you'd want. Frankly, what URL looks better:
http://commitfest.postgresql.org/action/commitfest_view?id=2
http://commitfest.postgresql.org/2009/07

Oh, and while at it, I don't foresee that we would want the commitfest
of july 2009 to change its name but not the "semantic" URL pointing to
its management. Or if it's ever wanted, as has been said, have a look
at Apache Rewrite Rules system, it's made for supporting content being
moved. Something in the spirit of:
RedirectPermanent /2009/07 /2009/08

While at it, I imagine that within a given commit fest, a single patch
will have a stable shortname, or if it comes to change, we could
accept that the URL too change:
http://commitfest.postgresql.org/action/patch_view?id=71
http://commitfest.postgresql.org/2009/07/Synch_Rep
http://commitfest.postgresql.org/current/Synch_Rep

Now, rather than following the names with apache setup, maybe you
could add something like some history tables tracking short name
changes (maybe some ON UPDATE trigger would do), so that referring to
an older name would send a HTTP 302 redirect to the new name URL?

I'd like your work to be useful for all of us and appreciated to its
real value, and those changes well seem like user interface
improvements rather than basic structure or behavior questioning.

Regards,
--
dim

> I *am* using some kind of key. Specifically, in integer derived from
> a serial column. It's just as stable as 16 random bytes displayed in
> hex, but a lot shorter and easier to remember, if you're the sort of
> person who likes to remember URLs. :-)
>

Wasn't aware of exately what you were doing. It sounded like multiple
things were in the query_string. If its already a single key, than
there is no need to use a different key. And no, I don't like
remebering URLs ... thus all the fuss about breaking bookmarks ;-)

>>>> It's impossible to know that this is commitfest 2009-07.
>>>>
>>> commitfest.postgresql.org/2009/07 ?
>>>
>>> That, or any similar scheme, seems easily doable with a
>>> little apache rewrite magic and some programming. See my
>>> blog urls for one such example.
>> IMHO, I don't see much gain to encoding the date into the url either. This
>> is not a great way of telling the user when something occurred. A lookup is
>> going to occur either way, so why not get all data at once using a single
>> method?
>
> Sorry, I'm not following this part.

Using a URL to encode when something occurred was being offered as a
solution to know what commitfest it is. I'm not sure where your
confusion is?

--
Andrew Chernow
eSilo, LLC
every bit counts
http://www.esilo.com/

On Fri, Jul 3, 2009 at 3:08 PM, Dimitri Fontaine<dfontaine(at)hi-media(dot)com> wrote:
> Your software seems to be better than a wiki, but its potential users are
> expressing needs and bikescheding. I think you'd better accept both kind of
> changes as long as it's not making your life much harder than you'd want.
> Frankly, what URL looks better:
>  http://commitfest.postgresql.org/action/commitfest_view?id=2
>  http://commitfest.postgresql.org/2009/07

We've definitely gotten to the "harder than I'd want" point at this point.

> Oh, and while at it, I don't foresee that we would want the commitfest of
> july 2009 to change its name but not the "semantic" URL pointing to its
> management. Or if it's ever wanted, as has been said, have a look at Apache
> Rewrite Rules system, it's made for supporting content being moved.
> Something in the spirit of:
>  RedirectPermanent /2009/07 /2009/08

I humbly submit that it's insane to thing about editing httpd.conf
every time somebody renames an object. The point here is to *reduce*
the administrative burden of maintaining this thing.

> While at it, I imagine that within a given commit fest, a single patch will
> have a stable shortname, or if it comes to change, we could accept that the
> URL too change:
>  http://commitfest.postgresql.org/action/patch_view?id=71
>  http://commitfest.postgresql.org/2009/07/Synch_Rep
>  http://commitfest.postgresql.org/current/Synch_Rep
>
> Now, rather than following the names with apache setup, maybe you could add
> something like some history tables tracking short name changes (maybe some
> ON UPDATE trigger would do), so that referring to an older name would send a
> HTTP 302 redirect to the new name URL?

So, the good thing about the current system is that the URL *doesn't*
change and you *don't* need complicated bookkeeping to remember every
URL you've ever assigned to the thing to get it. I am well aware that
it's possible to design a system that does this, but what benefit do
we get out of it? Making it possible to tell what a URL points to
without clicking on it, for those occasions when you're stranded on a
desert island with a URL and no Internet access?

There's a subtle and pernicious danger of the system you're proposing,
too. If we regularly change the URLs that refer to certain objects,
but compensate for that by remembering the whole history of URLs that
have ever referred to that object, then there will multiple URLs that
refer to the same object, of which all but one will contain WRONG
information about what that link points to. The
http://commitfest.postgresql.org/2009/07/Sync_Rep link, for example,
might imply to someone that the patch is in the 2009-07 commitfest,
but in fact it may well have been moved to the 2009-11, 2010-01, or
2010-03 commitfest, or we might have finally come to our senses and
realized that it ought to be called "streaming rep" and rename it.

It's true that the current URLs, without some sort of accompanying
text, do not tell you what they point to. But no information can
often be better than disinformation.

> I'd like your work to be useful for all of us and appreciated to its real
> value, and those changes well seem like user interface improvements rather
> than basic structure or behavior questioning.

I'd like that, too.

...Robert

On Fri, Jul 3, 2009 at 3:15 PM, Andrew Chernow<ac(at)esilo(dot)com> wrote:
>> I *am* using some kind of key.  Specifically, in integer derived from
>> a serial column.  It's just as stable as 16 random bytes displayed in
>> hex, but a lot shorter and easier to remember, if you're the sort of
>> person who likes to remember URLs.  :-)
>>
>
> Wasn't aware of exately what you were doing.  It sounded like multiple
> things were in the query_string.  If its already a single key, than there is
> no need to use a different key.  And no, I don't like remebering URLs ...
> thus all the fuss about breaking bookmarks ;-)

Right. The current system has exactly ZERO chance of breaking any
bookmarks, and all of the proposed alternatives are much more likely
to do so.

>>>>> It's impossible to know that this is commitfest 2009-07.
>>>>>
>>>> commitfest.postgresql.org/2009/07 ?
>>>>
>>>> That, or any similar scheme, seems easily doable with a
>>>> little apache rewrite magic and some programming. See my
>>>> blog urls for one such example.
>>>
>>> IMHO, I don't see much gain to encoding the date into the url either.
>>> This
>>> is not a great way of telling the user when something occurred.  A lookup
>>> is
>>> going to occur either way, so why not get all data at once using a single
>>> method?
>>
>> Sorry, I'm not following this part.
>
> Using a URL to encode when something occurred was being offered as a
> solution to know what commitfest it is.  I'm not sure where your confusion
> is?

The suggestion was to encode the start date of the CommitFest in the
URL, instead of using a non-natural key.

...Robert

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> Backing up for a moment to ten thousand feet here, I posted a link to
> this web app on May 26th. I received several comments on it, all of
> them positive, including some constructive feedback from you which I
> took to heart. It is now July 1st, and I am trying very hard to get
> ready for the next CommitFest, which I have agreed to manage. So I
> need to determine whether there is some finite number of changes of
> manageable size that I can make to get this to a state where we can
> use it, or whether I should give up hope now and go back to the wiki.

I think it's probably fixable, if you've got some time to put into it
between now and the 15th. What's being griped about is user interface
details, and it's not surprising that you as the author didn't see
these things the same way a new user would. From what I've been able to
see, the underlying functionality is mostly there, but it needs some
usability/presentation tweaking.

> I accept the need for and am willing to make the following changes:

> - Changing the patch comment field from type text to type textarea and
> integrating it into the patch view page to provide context.
> - Adding a note to the effect that the message ID is optional.
> - Adding stable links with mnemonic names for the open, in progress,
> and most recently closed commitfests.

> With respect to the issue of the page URLs, I'm very unconvinced of
> the value of making a change.

Given your item 3 above, I think we can live with the URLs otherwise.

One other thing I was noticing is that the items for a particular patch
seem to be listed in reverse date order. Personally I find this strange
and would prefer newest-at-the-bottom --- in particular, having the
patch itself at the bottom doesn't seem especially usable. We might
need to take a vote on that though, since I suppose some people like
newest-at-the-top.

regards, tom lane

On Fri, Jul 3, 2009 at 3:27 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> Backing up for a moment to ten thousand feet here, I posted a link to
>> this web app on May 26th.  I received several comments on it, all of
>> them positive, including some constructive feedback from you which I
>> took to heart.  It is now July 1st, and I am trying very hard to get
>> ready for the next CommitFest, which I have agreed to manage.  So I
>> need to determine whether there is some finite number of changes of
>> manageable size that I can make to get this to a state where we can
>> use it, or whether I should give up hope now and go back to the wiki.
>
> I think it's probably fixable, if you've got some time to put into it
> between now and the 15th.  What's being griped about is user interface
> details, and it's not surprising that you as the author didn't see
> these things the same way a new user would.  From what I've been able to
> see, the underlying functionality is mostly there, but it needs some
> usability/presentation tweaking.

Thanks, that is really good news. I agree that the presentation and
usability need some work and I'm trying to address those concerns as
expediently as I can. Part of my angst is that I am going to have
only sporadic Internet access for the next week, so what I can't get
done today (while my wife wonders why I am doing a second job on a
holiday for no money) probably isn't going to happen for a bit, and I
would like to get cut over so that Brendan doesn't have to keep
manually replicating changes between the systems. I will see if I can
make the changes below happen today.

>> I accept the need for and am willing to make the following changes:
>
>> - Changing the patch comment field from type text to type textarea and
>> integrating it into the patch view page to provide context.
>> - Adding a note to the effect that the message ID is optional.
>> - Adding stable links with mnemonic names for the open, in progress,
>> and most recently closed commitfests.
>
>> With respect to the issue of the page URLs, I'm very unconvinced of
>> the value of making a change.
>
> Given your item 3 above, I think we can live with the URLs otherwise.

/me feels like he has dodged a bullet.

> One other thing I was noticing is that the items for a particular patch
> seem to be listed in reverse date order.  Personally I find this strange
> and would prefer newest-at-the-bottom --- in particular, having the
> patch itself at the bottom doesn't seem especially usable.  We might
> need to take a vote on that though, since I suppose some people like
> newest-at-the-top.

I think it IS newest at the bottom, and I agree that that is how it
should be. 24 hours ago it was alpha by topic and then alpha by patch
name, but now it is topic by sortorder, then topic by name, then patch
by ascending ID number (which works out to newest at the bottom). I
thought the other way would be OK, but after Brendan and I imported
the data we both said "that sucks", so it got changed last night
around midnight Eastern +/- an hour.

One of the things that I would like to add in the future is the
ability to assign a patch a shortname. This would be useful for
building command-line tools to interface with the system, e.g.
download the sepgsql patch. The idea would be that these names would
be stable, though of course it's hard to see how to guarantee that
100%. I am still trying to work out in my mind how best to set that
up, though, so it's probably not going to happen right away unless
someone else is prepared to do some of the legwork.

...Robert

Robert Haas <robertmhaas(at)gmail(dot)com> wrote:

> I think it IS newest at the bottom, and I agree that that is how it
> should be.

It seems to be inconsistent. Probably because everything wound up
with the same date, the order is probably more-or-less random. What
are the chances that the date or timestamp of the corresponding wiki
modification could be put onto the patch and comment lines? (One
would hope that could be done with a script, rather than by hand....)

-Kevin

"Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
> Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>> I think it IS newest at the bottom, and I agree that that is how it
>> should be.

> It seems to be inconsistent. Probably because everything wound up
> with the same date, the order is probably more-or-less random.

Yeah, I think that's what I'm seeing.

> What
> are the chances that the date or timestamp of the corresponding wiki
> modification could be put onto the patch and comment lines? (One
> would hope that could be done with a script, rather than by hand....)

Currently, it seems that most or all of the entries are links to
archived messages. Scraping the date from the underlying message
would be the best thing.

regards, tom lane

Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Currently, it seems that most or all of the entries are links to
> archived messages. Scraping the date from the underlying message
> would be the best thing.

Just for purposes of conversion, or as a long-term behavior?

-Kevin

"Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Currently, it seems that most or all of the entries are links to
>> archived messages. Scraping the date from the underlying message
>> would be the best thing.

> Just for purposes of conversion, or as a long-term behavior?

Well, right at the moment I was just thinking of the conversion problem,
but it wouldn't be a horrible idea to make it a permanent behavior for
entries that include a message reference.

regards, tom lane

On Friday 03 July 2009 07:57:35 Robert Haas wrote:
> We're still hacking on a few other details of the formatting and
> interface, but you might want to cruise over and have a look.

One thing that I noticed it that it takes too many clicks in general to make a
set of changes. When I look at a patch, why not have the controls to add a
comment and make changes right there, instead of on another page? For
instance, the functionality of

action/patch_view?id=69
action/patch_comment_form?patch=69
action/patch_form?id=69
action/patch_bump?id=69

could all be on the same page.

And please make "Delete Patch" into a button instead of a link.

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> And please make "Delete Patch" into a button instead of a link.

Only if there's some kind of confirmation ...

regards, tom lane

On Fri, 2009-07-03 at 16:54 -0400, Tom Lane wrote:
> Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> > And please make "Delete Patch" into a button instead of a link.
>
> Only if there's some kind of confirmation ...

Should we actually delete patches? I get removing them from the list but
it seems there could be benefit from keeping patches that are not
"quite" there or perhaps present an idea that is ahead of its time (or
half baked but interesting in genera?).

+1 on the confirmation regardless.

Joshua D. Drake

>
> regards, tom lane
>
--
PostgreSQL - XMPP: jdrake(at)jabber(dot)postgresql(dot)org
Consulting, Development, Support, Training
503-667-4564 - http://www.commandprompt.com/
The PostgreSQL Company, serving since 1997

On Fri, Jul 3, 2009 at 5:49 PM, Joshua D. Drake<jd(at)commandprompt(dot)com> wrote:
> On Fri, 2009-07-03 at 16:54 -0400, Tom Lane wrote:
>> Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
>> > And please make "Delete Patch" into a button instead of a link.
>>
>> Only if there's some kind of confirmation ...
>
> Should we actually delete patches? I get removing them from the list but
> it seems there could be benefit from keeping patches that are not
> "quite" there or perhaps present an idea that is ahead of its time (or
> half baked but interesting in genera?).
>
> +1 on the confirmation regardless.

There is a confirmation right now, plus it doesn't work at all unless
you delete all the comments first, which you can't do either unless
you are an administrator or the only person who has ever commented on
the patch.

Really, deleting patches should be quite rare and only needed in cases
where the patch should not have been added in the first place.

I guess I'm not really seeing why that particular thing should be a
button rather than a link. It would mess up the formatting for no
obvious benefit.

...Robert

On Fri, 2009-07-03 at 17:57 -0400, Robert Haas wrote:

> I guess I'm not really seeing why that particular thing should be a
> button rather than a link. It would mess up the formatting for no
> obvious benefit.
>

Not arguing one way or the other, a button says, "I am about to perform
X". A link *always* says, "I am about to go to a new web page".

Joshua D. Drake

> ...Robert
>
--
PostgreSQL - XMPP: jdrake(at)jabber(dot)postgresql(dot)org
Consulting, Development, Support, Training
503-667-4564 - http://www.commandprompt.com/
The PostgreSQL Company, serving since 1997

On Fri, Jul 3, 2009 at 6:19 PM, Joshua D. Drake<jd(at)commandprompt(dot)com> wrote:
> On Fri, 2009-07-03 at 17:57 -0400, Robert Haas wrote:
>
>> I guess I'm not really seeing why that particular thing should be a
>> button rather than a link.  It would mess up the formatting for no
>> obvious benefit.
>>
>
> Not arguing one way or the other, a button says, "I am about to perform
> X". A link *always* says, "I am about to go to a new web page".

Hmm, there is some truth to what you say. I guess the way I think
about it, a button says "I am about to submit this form" and a link
says "I am about to do something other than submit a form". But it's
certainly an arguable point.

...Robert

On Fri, Jul 3, 2009 at 3:40 PM, Robert Haas<robertmhaas(at)gmail(dot)com> wrote:
>>> I accept the need for and am willing to make the following changes:
>>
>>> - Changing the patch comment field from type text to type textarea and
>>> integrating it into the patch view page to provide context.
>>> - Adding a note to the effect that the message ID is optional.
>>> - Adding stable links with mnemonic names for the open, in progress,
>>> and most recently closed commitfests.

Done.
Done.
Done.

You guys are harder to please than my boss. He doesn't quite know for
sure which things are possible. :-)

...Robert

On Fri, Jul 3, 2009 at 4:43 PM, Peter Eisentraut<peter_e(at)gmx(dot)net> wrote:
> On Friday 03 July 2009 07:57:35 Robert Haas wrote:
>> We're still hacking on a few other details of the formatting and
>> interface, but you might want to cruise over and have a look.
>
> One thing that I noticed it that it takes too many clicks in general to make a
> set of changes.  When I look at a patch, why not have the controls to add a
> comment and make changes right there, instead of on another page?  For
> instance, the functionality of
>
> action/patch_view?id=69
> action/patch_comment_form?patch=69
> action/patch_form?id=69
> action/patch_bump?id=69
>
> could all be on the same page.

Well, I've added the comment form to that page, by request, but I
don't see how you could fit the rest on there in any sort of sane way.
But I'm accepting patches...

...Robert

On Fri, Jul 3, 2009 at 4:00 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
>> Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>>> I think it IS newest at the bottom, and I agree that that is how it
>>> should be.
>
>> It seems to be inconsistent.  Probably because everything wound up
>> with the same date, the order is probably more-or-less random.
>
> Yeah, I think that's what I'm seeing.

I think what you guys must be seeing is that the topics are not in
quite the same order. As far as I can tell, the ordering of patches
within topics is absolutely identical. If you see a counterexample,
please point it out. If you want to fix the order of the topics,
click on the "CommitFest Topics" link in the upper lefthand corner of
the page and frobnicate the sort order.

>> What
>> are the chances that the date or timestamp of the corresponding wiki
>> modification could be put onto the patch and comment lines?  (One
>> would hope that could be done with a script, rather than by hand....)
>
> Currently, it seems that most or all of the entries are links to
> archived messages.  Scraping the date from the underlying message
> would be the best thing.

Brendan, is this something that you can work on? Or does anyone else
have some time to put into it? And do we have to fix this before we
can declare the new system live, or can we catch up after the fact?
Because replicating changes from the old system to the new is a bit of
a pain in the neck.

...Robert

2009/7/4 Robert Haas <robertmhaas(at)gmail(dot)com>:
> On Fri, Jul 3, 2009 at 4:00 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
>>> What
>>> are the chances that the date or timestamp of the corresponding wiki
>>> modification could be put onto the patch and comment lines?  (One
>>> would hope that could be done with a script, rather than by hand....)
>>
>> Currently, it seems that most or all of the entries are links to
>> archived messages.  Scraping the date from the underlying message
>> would be the best thing.
>
> Brendan, is this something that you can work on?

I will take a stab at it. I think Tom's suggestion of harvesting the
time of the mail message from the archives could do the job.

Cheers,
BJ

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Fri, Jul 3, 2009 at 4:00 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
>>> It seems to be inconsistent. Probably because everything wound up
>>> with the same date, the order is probably more-or-less random.
>>
>> Yeah, I think that's what I'm seeing.

> I think what you guys must be seeing is that the topics are not in
> quite the same order. As far as I can tell, the ordering of patches
> within topics is absolutely identical.

No, what we're complaining about is the ordering of comments for a
single patch. If you look at
http://commitfest.postgresql.org/action/commitfest_view?id=2
in some cases there are Comments before the Patch itself, and in
some cases after, but more often than not the original Patch is
at the bottom.

regards, tom lane

2009/7/4 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> No, what we're complaining about is the ordering of comments for a
> single patch.

Now moot, since I've successfully pulled the dates for all comments
with a message-id from the archives, and updated the database
accordingly.

Cheers,
BJ

On Jul 3, 2009, at 11:56 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Fri, Jul 3, 2009 at 4:00 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
>>>> It seems to be inconsistent. Probably because everything wound up
>>>> with the same date, the order is probably more-or-less random.
>>>
>>> Yeah, I think that's what I'm seeing.
>
>> I think what you guys must be seeing is that the topics are not in
>> quite the same order. As far as I can tell, the ordering of patches
>> within topics is absolutely identical.
>
> No, what we're complaining about is the ordering of comments for a
> single patch. If you look at
> http://commitfest.postgresql.org/action/commitfest_view?id=2
> in some cases there are Comments before the Patch itself, and in
> some cases after, but more often than not the original Patch is
> at the bottom.

Oh, duh. I totally missed that.

...Robert

On Saturday 04 July 2009 01:19:23 Joshua D. Drake wrote:
> On Fri, 2009-07-03 at 17:57 -0400, Robert Haas wrote:
> > I guess I'm not really seeing why that particular thing should be a
> > button rather than a link. It would mess up the formatting for no
> > obvious benefit.
>
> Not arguing one way or the other, a button says, "I am about to perform
> X". A link *always* says, "I am about to go to a new web page".

That was my feeling.

Le 5 juil. 09 à 00:13, Peter Eisentraut a écrit :
> On Saturday 04 July 2009 01:19:23 Joshua D. Drake wrote:
>> Not arguing one way or the other, a button says, "I am about to
>> perform
>> X". A link *always* says, "I am about to go to a new web page".
>
> That was my feeling.

And bots (google etc) will follow links.
--
dim

On Jul 4, 2009, at 5:18 PM, Dimitri Fontaine <dfontaine(at)hi-media(dot)com>
wrote:

> Le 5 juil. 09 à 00:13, Peter Eisentraut a écrit :
>> On Saturday 04 July 2009 01:19:23 Joshua D. Drake wrote:
>>> Not arguing one way or the other, a button says, "I am about to
>>> perform
>>> X". A link *always* says, "I am about to go to a new web page".
>>
>> That was my feeling.
>
> And bots (google etc) will follow links.

They won't log in, though. :-)

Someone want to write a patch, then?

..Robert

2009/7/3 Robert Haas <robertmhaas(at)gmail(dot)com>:
> The application stamps comments with the
> community login of the person who left them, but the import stamped
> them with names instead.  This is actually of some significance, since
> the app will allow you to edit your own comments but not those of
> other people.  We could probably fix this if someone can give us
> access to (or a dump of) the realname to username mappings from the
> community login DB.

Nobody came forward with a mapping of real names to community logins,
so I went ahead and did this the dumb, slow way (eyeballing the wiki
history and manually creating a mapping for the names we have in the
commitfest app so far).

I've updated the patch comment creator field with the login names I
was able to map, but there are a few contributors who either don't
have a community login, or haven't used the wiki. In those cases I
left the name as-is.

You should now be able to edit comments that you created. If you
think you should be able to edit a comment, but you can't because the
login name is wrong, let me know and I'll fix it up.

Cheers,
BJ

Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:
> On Saturday 04 July 2009 01:19:23 Joshua D. Drake wrote:
>> a button says, "I am about to perform X".
>> A link *always* says, "I am about to go to a new web page".
>
> That was my feeling.

In addition, if the action will be preceded by a dialog (for options
or confirmation) the button text should end with '...'.

-Kevin

Hi folks,

We're now about a week away from the start of the July 2009
commitfest, and we need to make a decision about whether to start
using http://commitfest.postgresql.org to manage it, or punt to the
next commitfest and continue to use the wiki for July.

Robert and I have been upgrading the app in response to the feedback
so far, and personally I think the app is already a far superior tool
to the wiki for managing a commitfest. I think we should use it for
2009-07.

What we need in order to proceed is:
a) a viable consensus from patch reviewers/committers to switch over,
b) an email out to -hackers about the new app,
c) updates to the wiki developer pages pointing to the new app, and
d) decommissioning the wiki July CF.

I can do b), c) and d) but I could use your help in obtaining a).

If you think we should switch over, please say so.

If you think the app needs some tweaking, please say so but let's
switch over anyway and make those tweaks as we go.

If you think the app is fundamentally less useful than the wiki,
please say so and we'll work out whether we can resolve your objection
in time for the start of the CF.

Thanks in advance for your comments.

Cheers,
BJ

Brendan Jurd <direvus(at)gmail(dot)com> wrote:

> If you think the app is fundamentally less useful than the wiki,
> please say so and we'll work out whether we can resolve your
> objection in time for the start of the CF.

It's been down for a while now. I don't know if this is causal, but
the failure seemed to start when I went into the patch I submitted,
and clicked on the link to edit a comment I added.

At any rate, the reliability doesn't seem to match the wiki yet, which
would seem to be a pretty fundamental thing to fix before switching.

-Kevin

Brendan Jurd <direvus(at)gmail(dot)com> wrote:

> If you think the app is fundamentally less useful than the wiki,
> please say so and we'll work out whether we can resolve your
> objection in time for the start of the CF.

Oh, sure -- I post about it being down, and seconds after I hit send
it comes up again. :-/

Do we know that cause?

-Kevin

2009/7/8 Kevin Grittner <Kevin(dot)Grittner(at)wicourts(dot)gov>:
> Oh, sure -- I post about it being down, and seconds after I hit send
> it comes up again.   :-/
>
> Do we know that cause?

Well, no, since I've never observed it being "down" and I really have
no idea what you mean by that.

Maybe you could describe the symptoms you observed? Was the webserver
totally uncontactable, or was it an error in the web app itself?

Cheers,
BJ

Brendan Jurd <direvus(at)gmail(dot)com> wrote:

> Maybe you could describe the symptoms you observed? Was the
> webserver totally uncontactable, or was it an error in the web app
> itself?

When I clicked the link to edit the comment, it clocked until the
browser timed out. So then I tried the URL for the main page, and it
again clocked until the browser timed out. I then tried again on the
main page, and it clocked for a minute, so I posted the message about
it being down. Moments after I sent that, the page came up.

If anyone else was hitting the site and it was working, perhaps the
problem could have been elsewhere. (Our network or our Internet path
to the site could have had some temporary issue.)

-Kevin

On Tuesday 07 July 2009 11:29:07 Brendan Jurd wrote:
> We're now about a week away from the start of the July 2009
> commitfest, and we need to make a decision about whether to start
> using http://commitfest.postgresql.org to manage it, or punt to the
> next commitfest and continue to use the wiki for July.

I have the following concern: Likely, this tool and the overall process will
evolve over time. To pick an example that may or may not be actually useful,
in the future we might want to change from a fixed list of patch sections to a
free list of tags, say. Then someone might alter the application backend, and
we'd use that new version for the next commit fest at the time. What will
that do to the data of old commit fests?

With the wiki, the data of the old fests will pretty much stay what is was,
unless we change the wiki templates in drastic ways, as I understand it. But
if we did changes like the above, or more complicated things, perhaps, what
will happen? Perhaps we simply don't care about the historical data. But if
we do, we better have pretty high confidence that the current application will
do for a while or is easily upgradable.

Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:

> in the future we might want to change from a fixed list of patch
> sections to a free list of tags, say. Then someone might alter the
> application backend, and we'd use that new version for the next
> commit fest at the time. What will that do to the data of old
> commit fests?

Certainly you see how trivial that conversion would be. If that were
the worst case anyone could even imagine, it would be a pretty strong
argument that the schema is more than good enough to proceed.

Do you see anything fundamentally wrong with the structure in terms
of long term goals?

-Kevin

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> With the wiki, the data of the old fests will pretty much stay what is
> was, unless we change the wiki templates in drastic ways, as I
> understand it. But if we did changes like the above, or more
> complicated things, perhaps, what will happen? Perhaps we simply
> don't care about the historical data. But if we do, we better have
> pretty high confidence that the current application will do for a
> while or is easily upgradable.

I'm not convinced that we care in the least about commitfests that are
more than a fest or two back; especially since the mailing lists archive
all the interesting underlying data. However, if we did, the answer
doesn't seem that hard: keep the old database instance on-line for
serving up the old data, and put a new one beside it.

regards, tom lane

On Jul 7, 2009, at 2:14 PM, Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:

> On Tuesday 07 July 2009 11:29:07 Brendan Jurd wrote:
>> We're now about a week away from the start of the July 2009
>> commitfest, and we need to make a decision about whether to start
>> using http://commitfest.postgresql.org to manage it, or punt to the
>> next commitfest and continue to use the wiki for July.
>
> I have the following concern: Likely, this tool and the overall
> process will
> evolve over time. To pick an example that may or may not be
> actually useful,
> in the future we might want to change from a fixed list of patch
> sections to a
> free list of tags, say. Then someone might alter the application
> backend, and
> we'd use that new version for the next commit fest at the time.
> What will
> that do to the data of old commit fests?

I don't see this as being much of an obstacle. I have migrated data
far more complex, and I venture to say that most of the rest of the
regular posters in this forum have too.

> With the wiki, the data of the old fests will pretty much stay what
> is was,
> unless we change the wiki templates in drastic ways, as I understand
> it. But
> if we did changes like the above, or more complicated things,
> perhaps, what
> will happen? Perhaps we simply don't care about the historical
> data. But if
> we do, we better have pretty high confidence that the current
> application will
> do for a while or is easily upgradable.

I suspect both are true, but in the unlikely event that we decide on
some massive change to the system, we can either run the DBs in
parallel as Tom suggests, or dump out the older data in Wiki markup
and post it on there. But I can't imagine what we'd want to do that
would even make us consider such drastic steps. Your example would
not be a difficult migration, for instance.

...Robert

Robert Haas escribió:

> I suspect both are true, but in the unlikely event that we decide on
> some massive change to the system, we can either run the DBs in parallel
> as Tom suggests, or dump out the older data in Wiki markup and post it on
> there. But I can't imagine what we'd want to do that would even make us
> consider such drastic steps. Your example would not be a difficult
> migration, for instance.

By the way, if the migration of the current commitfest was an automatic
procedure, is there a chance that the old commitfests can be migrated as
well?

--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

2009/7/8 Alvaro Herrera <alvherre(at)commandprompt(dot)com>:
>
> By the way, if the migration of the current commitfest was an automatic
> procedure, is there a chance that the old commitfests can be migrated as
> well?
>

It wasn't really automatic as such. I used a few scripts that I saved
in case we needed to use them again, but there was also quite a bit of
manual massage of the data required in between those script runs.
There are a few minor incompatibilities, for example the wiki allows
you to put links to the mailing list archives anywhere in a review or
comment with {{messageLink}} but the app has the message-id as a
separate field. I like the way the data is laid out in the app
better, but it makes it difficult to migrate comments with multiple
{{messageLinks}} in them.

Short answer: I could bring across the old commitfests but it would
take a couple hours at best per commitfest and result in little bits
of data loss here and there. I think we might be better off just
leaving the closed commitfests up on the wiki, and putting a notice on
the app saying "commitfests prior to July 2009 can be found at
wiki.postgresql.org".

Cheers,
BJ

2009/7/8 Peter Eisentraut <peter_e(at)gmx(dot)net>:
> I have the following concern: Likely, this tool and the overall process will
> evolve over time.  To pick an example that may or may not be actually useful,
> in the future we might want to change from a fixed list of patch sections to a
> free list of tags, say.  Then someone might alter the application backend, and
> we'd use that new version for the next commit fest at the time.  What will
> that do to the data of old commit fests?

Same thing that always happens when you make big-picture schema
changes to an app ... you either migrate the data or you leave a
historical instance in place.

NB moving from "topics" to "tags" is *not* a big-picture change and
wouldn't invalidate past data in the slightest.

> With the wiki, the data of the old fests will pretty much stay what is was,
> unless we change the wiki templates in drastic ways, as I understand it.

Actually the wiki makes this more difficult. Changes to the templates
are limited to adding new keyword arguments. If you want to, say,
insert a new positional argument into one of the templates, you need
to make a copy of all the affected templates and migrate all existing
template calls to the historical copy. We had to do this once. It
wasn't pleasant.

So, even for minor changes to layout (like the "tags" thing you
mentioned) result in having to leave a historical instance of the
system in place. Whereas having the data in a relational database
makes it far more likely that we can just migrate across small,
incremental changes.

Cheers,
BJ

robertmhaas(at)gmail(dot)com (Robert Haas) writes:
> I suspect both are true, but in the unlikely event that we decide on
> some massive change to the system, we can either run the DBs in
> parallel as Tom suggests, or dump out the older data in Wiki markup
> and post it on there. But I can't imagine what we'd want to do that
> would even make us consider such drastic steps. Your example would
> not be a difficult migration, for instance.

We had an ancient version of Bugzilla in use for quite a while; it was
*SO* vastly different from modern versions that it wasn't remotely
plausible to port the data from the old instance to a new one.

I went and ran "wget" to pull all the contents of the old instance,
turning that into a series of static web pages. No longer updatable,
but certainly browsable.

Once a CommitFest is complete, I could easily see making a summary of
it, as a series of static web pages. No need for a database anymore
altogether ;-).
--
select 'cbbrowne' || '@' || 'linuxfinances.info';
http://linuxdatabases.info/info/spreadsheets.html
I'd give my right arm to be ambidextrous!

Brendan Jurd escribió:

> Short answer: I could bring across the old commitfests but it would
> take a couple hours at best per commitfest and result in little bits
> of data loss here and there. I think we might be better off just
> leaving the closed commitfests up on the wiki, and putting a notice on
> the app saying "commitfests prior to July 2009 can be found at
> wiki.postgresql.org".

Agreed; if it requires manual work let's leave them in the wiki.

--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

Brendan Jurd <direvus(at)gmail(dot)com> writes:
> We're now about a week away from the start of the July 2009
> commitfest, and we need to make a decision about whether to start
> using http://commitfest.postgresql.org to manage it, or punt to the
> next commitfest and continue to use the wiki for July.

While reorganizing my bookmarks for this I realized that there is a
fairly significant bit of functionality that's entirely missing from
the new app. With the wiki page, you could conveniently see what had
been done lately by examining the page history. I don't see any
equivalent capability in the new app. I find this fairly significant,
as evidenced by the fact that I'd gone so far as to set up a bookmark
for the history view. I'm not particularly wedded to the wiki page
history in terms of what it looks like or how it functions, but I do
feel a need to know what other people have done recently.

regards, tom lane

Em Thu, 09 Jul 2009 14:16:41 -0300, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> escreveu:
> I'm not particularly wedded to the wiki page
> history in terms of what it looks like or how it functions, but I do
> feel a need to know what other people have done recently.

May be the new app could have a page with a filterable table log with some
important columns like "who" do "what" on "where" and "when".

This could help, maybe with a RSS in that (like in git).

[]s
Dickson S. Guedes
http://pgcon.postgresql.org.br
http://www.postgresql.org.br

2009/7/10 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> While reorganizing my bookmarks for this I realized that there is a
> fairly significant bit of functionality that's entirely missing from
> the new app.  With the wiki page, you could conveniently see what had
> been done lately by examining the page history.  I don't see any
> equivalent capability in the new app.

You're right, the app currently lacks a view for this.

And while it would be pretty easy to add a page that just lists the
most recent comments in descending creation order, that would not show
things like

* patches being committed/rejected/punted;
* patches changed to a different subsection;
* changes to a patch's short name;
* edits to existing comments.

We don't AFAIK collect data about these events. However, we could
have certain actions trigger the creation of an automated comment
(e.g., "Status changed to Committed by petere") and let the
aforementioned comment view suffice for a "history".

Cheers,
BJ

On Thu, Jul 09, 2009 at 02:35:04PM -0300, Dickson S. Guedes wrote:
> This could help, maybe with a RSS in that (like in git).

+1 for the RSS feed, if only because I think it sounds neat :)

--
Joshua Tolley / eggyknap
End Point Corporation
http://www.endpoint.com

On Jul 9, 2009, at 12:16 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Brendan Jurd <direvus(at)gmail(dot)com> writes:
>> We're now about a week away from the start of the July 2009
>> commitfest, and we need to make a decision about whether to start
>> using http://commitfest.postgresql.org to manage it, or punt to the
>> next commitfest and continue to use the wiki for July.
>
> While reorganizing my bookmarks for this I realized that there is a
> fairly significant bit of functionality that's entirely missing from
> the new app. With the wiki page, you could conveniently see what had
> been done lately by examining the page history. I don't see any
> equivalent capability in the new app. I find this fairly significant,
> as evidenced by the fact that I'd gone so far as to set up a bookmark
> for the history view. I'm not particularly wedded to the wiki page
> history in terms of what it looks like or how it functions, but I do
> feel a need to know what other people have done recently

I'll fix this. Give me a couple days; my Internet access here at the
family vacation spot is not compatible with "git push".

...Robert

> We don't AFAIK collect data about these events. However, we could
> have certain actions trigger the creation of an automated comment
> (e.g., "Status changed to Committed by petere") and let the
> aforementioned comment view suffice for a "history".

Can you put in a simple event-recording trigger for all changes? We can
dress it up for easy viewing later, but if the data isn't collected, it
will be impossible to recreate.

--
Josh Berkus
PostgreSQL Experts Inc.
www.pgexperts.com

Josh Berkus <josh(at)agliodbs(dot)com> writes:
>> We don't AFAIK collect data about these events. However, we could
>> have certain actions trigger the creation of an automated comment
>> (e.g., "Status changed to Committed by petere") and let the
>> aforementioned comment view suffice for a "history".

> Can you put in a simple event-recording trigger for all changes?

+1. Just add an event log table. I think we'll wish we had one later.

regards, tom lane

2009/7/10 Josh Berkus <josh(at)agliodbs(dot)com>:
>
>> We don't AFAIK collect data about these events.  However, we could
>> have certain actions trigger the creation of an automated comment
>> (e.g., "Status changed to Committed by petere") and let the
>> aforementioned comment view suffice for a "history".
>
> Can you put in a simple event-recording trigger for all changes?  We can
> dress it up for easy viewing later, but if the data isn't collected, it will
> be impossible to recreate.
>

I could, but a db trigger would have no awareness of which user made
the change. That would make the information substantially less useful
IMO.

I've got access to the database but I'm not in a position to make
changes to the app frontend. I'm therefore inclined to wait the
prophesied couple days for Robert's proper fix for this problem than
dropping in a trigger which will only tell us part of the story.

Bear in mind that CF activity over the past week has been in the realm
of 0-4 changes per 24 hours, so it's not like we are talking about
huge amounts of throughput here. Making up the bulk of those changes
were new patches coming in (which involves adding a comment so the
change is already tracked) and patches getting committed (which is
stored in -hackers and version control anyway).

Cheers,
BJ

Brendan Jurd <direvus(at)gmail(dot)com> writes:
> Bear in mind that CF activity over the past week has been in the realm
> of 0-4 changes per 24 hours, so it's not like we are talking about
> huge amounts of throughput here.

Well, it'll be more once commitfest actually starts, but in any case
it's not going to be enough to make a log table be a resource hog.

regards, tom lane

On Jul 9, 2009, at 12:35 PM, Brendan Jurd wrote:
> We don't AFAIK collect data about these events. However, we could
> have certain actions trigger the creation of an automated comment
> (e.g., "Status changed to Committed by petere") and let the
> aforementioned comment view suffice for a "history".

Our main system at work does that; any kind of status is stored as a
raw, text "note". It sucks. It makes trying to query for specific
kinds of events difficult, and it wastes a bunch of space.

It's a lot better to record machine-readable information for machine-
created events. If you want to present it all as one, I suggest a
union view that turns the machine-understood data into a human-
understandable text format.
--
Decibel!, aka Jim C. Nasby, Database Architect decibel(at)decibel(dot)org
Give your computer some brain candy! www.distributed.net Team #1828

>> I think we might be better off just
>> leaving the closed commitfests up on the wiki, and putting a notice on
>> the app saying "commitfests prior to July 2009 can be found at
>> wiki.postgresql.org".

+1. That's why we're switching technogies at the beginning of a dev cycle.

--
Josh Berkus
PostgreSQL Experts Inc.
www.pgexperts.com

On Thu, Jul 9, 2009 at 2:20 PM, Robert Haas<robertmhaas(at)gmail(dot)com> wrote:
> On Jul 9, 2009, at 12:16 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
>> Brendan Jurd <direvus(at)gmail(dot)com> writes:
>>>
>>> We're now about a week away from the start of the July 2009
>>> commitfest, and we need to make a decision about whether to start
>>> using http://commitfest.postgresql.org to manage it, or punt to the
>>> next commitfest and continue to use the wiki for July.
>>
>> While reorganizing my bookmarks for this I realized that there is a
>> fairly significant bit of functionality that's entirely missing from
>> the new app.  With the wiki page, you could conveniently see what had
>> been done lately by examining the page history.  I don't see any
>> equivalent capability in the new app.  I find this fairly significant,
>> as evidenced by the fact that I'd gone so far as to set up a bookmark
>> for the history view.  I'm not particularly wedded to the wiki page
>> history in terms of what it looks like or how it functions, but I do
>> feel a need to know what other people have done recently
>
> I'll fix this. Give me a couple days; my Internet access here at the family
> vacation spot is not compatible with "git push".

OK, this is a little bit quick-and-dirty, and I'm sure I'll get some,
ah, gentle suggestions for improvement, but I've added an activity log
to the app.

...Robert