7.4 features list

When we started on 7.4 development, I was hoping for three major
features:

o PITR (point-in-time recovery)
o Win32 port
o Master-slave Postgres-R

Now, the last one isn't going to happen because we didn't get any
funding for the replication guys. The first one is being worked on by
Patrick at Red Hat, and I have seen a patch, but I am not sure how close
it is to completion. Patrick?

For Win32, I was moving along well until I hit the fork/exec changes.
They are extensive, and I am having problems because though the port is
good from PeerDirect, it does take more work to get it up to our coding
standards, and second, it is against 7.2, and I have to factor all
changes up to our current CVS code.

I am continuing on the work, but I am now uncertain I can complete by
code freeze in Mid-May, or even by the end of May.

I am unsure what we should do:

delay code freeze
allow Win32 patches during beta
delay Win32 for 7.5

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian wrote:
> When we started on 7.4 development, I was hoping for three major
> features:
>
> o PITR (point-in-time recovery)
> o Win32 port
> o Master-slave Postgres-R
>
> Now, the last one isn't going to happen because we didn't get any
> funding for the replication guys. The first one is being worked on by
> Patrick at Red Hat, and I have seen a patch, but I am not sure how close
> it is to completion. Patrick?

I am not near completion. However, I did post the remaining issues to
the hacker list last week. At the moment, there does not seem to be
negative feedback with the proposed modifications so I'll take that as
acceptance.

At the moment, I'm looking at a staggered integration at best. I'd
like to get the on-disk changes into cvs asap. The rest can be pieced
in over the next month or so. It's not the best solution but it is
manageable.

Patrick

On Mon, 5 May 2003, Bruce Momjian wrote:

> When we started on 7.4 development, I was hoping for three major
> features:
>
> o PITR (point-in-time recovery)
> o Win32 port
> o Master-slave Postgres-R
>
> Now, the last one isn't going to happen because we didn't get any
> funding for the replication guys. The first one is being worked on by
> Patrick at Red Hat, and I have seen a patch, but I am not sure how close
> it is to completion. Patrick?
>
> For Win32, I was moving along well until I hit the fork/exec changes.
> They are extensive, and I am having problems because though the port is
> good from PeerDirect, it does take more work to get it up to our coding
> standards, and second, it is against 7.2, and I have to factor all
> changes up to our current CVS code.
>
> I am continuing on the work, but I am now uncertain I can complete by
> code freeze in Mid-May, or even by the end of May.
>
> I am unsure what we should do:
>
> delay code freeze
> allow Win32 patches during beta
> delay Win32 for 7.5

I've wondered recently at other mentions of the impending feature freeze and
wondering why that appears to have been set in stone. I know from comments
leading to 7.3 that there have been problems in the past with timetable
slippage but is it really sensible to essentially throw out a major feature
that will take another 6+ months to be released for the sake of a short delay
for the coming release?

Considering the slippage danger I'd say getting realistic estimates of the
timetable for these features would be a good idea. Freezing can then be fixed
at a time where there is a good chance the code will be done by and with the
'feature freeze' part emphasised, i.e. let the timetable slip this once but
that's it.

--
Nigel J. Andrews

Bruce Momjian schrieb:
> I am continuing on the work, but I am now uncertain I can complete by
> code freeze in Mid-May, or even by the end of May.
>
> I am unsure what we should do:
>
> delay code freeze
> allow Win32 patches during beta
> delay Win32 for 7.5
>
I'm just a "normal" user, but I'd like to throw in my to 2cents as well. As
much as I appreciate all the work you are doing, I think that decent Win32
version will give PostgreSQL big momentum especially over MySQL. I'm
happily running the Peer-Direct version (although not in a
business-critical environment) and I would just love to see an "official"
win32 version.

So from my point of view delaying Win32 for 7.5 would really be a bad choice.

Cheers
Thomas

"Nigel J. Andrews" <nandrews(at)investsystems(dot)co(dot)uk> writes:
> I've wondered recently at other mentions of the impending feature freeze and
> wondering why that appears to have been set in stone. I know from comments
> leading to 7.3 that there have been problems in the past with timetable
> slippage but is it really sensible to essentially throw out a major feature
> that will take another 6+ months to be released for the sake of a short delay
> for the coming release?

Unfortunately, we've heard that siren song before :-(

"Hold up this release just a little while I finish major feature X" was
exactly the logic that led us to slip the 7.1 and 7.2 releases by many
months, because "just a little" kept stretching. I don't think it's
fair to other developers to put the release on ice while one or two
people finish what they were supposed to have done on time. No one else
can plan to do useful work if the release schedule is driven by
"whenever Bruce/Patrick/Vadim/Tom/whoever gets done with X".

If we're going to slip then we should establish a new release target
date several months away (September or later), and hold to that schedule.
But is that really a better plan? We have accomplished quite a lot
since last September, and holding those improvements out of the hands of
users for months more is not a zero-cost thing. Nor do I take it as a
sure thing that we won't be hearing the very same complaint come
September. Both of these projects were supposed to have been done
months ago, and they're not done. That gives me no confidence that
they'll meet the next deadline we set. If September comes and we still
have no working Win32 port, will you vote to delay again?

regards, tom lane

Hi there,

On Mon, 5 May 2003, Thomas Kellerer wrote:

> Bruce Momjian schrieb:

> > I am continuing on the work, but I am now uncertain I can complete by
> > code freeze in Mid-May, or even by the end of May.
> >
> > I am unsure what we should do:
> >
> > delay code freeze
> > allow Win32 patches during beta
> > delay Win32 for 7.5
> >

> I'm just a "normal" user, but I'd like to throw in my to 2cents as well. As
> much as I appreciate all the work you are doing, I think that decent Win32
> version will give PostgreSQL big momentum especially over MySQL. I'm
> happily running the Peer-Direct version (although not in a
> business-critical environment) and I would just love to see an "official"
> win32 version.
>
> So from my point of view delaying Win32 for 7.5 would really be a bad choice.

That's also my opinion. Sometimes Justin forwards mails he
receives from people concerned in PostgreSQL to the advocacy mailing list,
and you'll notice that most of that people look like non-technical (ie,
managers) and probably would like to develop their software without
worrying about the plataform their database will have to run.
Unfortunatelly I currently don't track -current nor have yet tested a
Win32 snapshot (though I'd like to, once I looked for but I couldn't find
any binary snapshot).

If there's something to improve a lot the advocacy effort results,
this would be a Win32 port.

[]s
Ricardo.

On Mon, 5 May 2003, Tom Lane wrote:

> ...
>
> If we're going to slip then we should establish a new release target
> date several months away (September or later), and hold to that schedule.

That's why I say try and get a realistic estimate and then a judgment can be
made as to whether the freeze stays or it slips _a little_.

> But is that really a better plan? We have accomplished quite a lot
> since last September, and holding those improvements out of the hands of
> users for months more is not a zero-cost thing. Nor do I take it as a
> sure thing that we won't be hearing the very same complaint come
> September. Both of these projects were supposed to have been done
> months ago, and they're not done. That gives me no confidence that
> they'll meet the next deadline we set. If September comes and we still
> have no working Win32 port, will you vote to delay again?

No, I wouldn't, but then I was thinking more of a month, two at the very most,
certainly not stretching it as far as Sept.

For a current client I actually prefer to get an early release, even June is
pushing it a bit. On the other hand if the release isn't going to be out in
time for me to switch to it for the first release of this project, which it may
well not be on the current release schedule, then I'd certainly like one with
PITR available to make upgrading attractive not just done for the sake of
upgrading.

--
Nigel J. Andrews

Amen.

Since I now have some free time on my hands (through no choice of mine
;-( ) I can devote some of it to the W32 work if someone (Bruce?) would
like to throw some tasks over the fence at me, to help shorten the cycle. (I
take it from previous discussion what is basically needed is Windows+cygwin,
right?)

andrew

----- Original Message -----
From: "Thomas Kellerer" <spam_eater(at)gmx(dot)net>
To: <pgsql-hackers(at)postgresql(dot)org>
Sent: Monday, May 05, 2003 5:02 PM
Subject: Re: [HACKERS] 7.4 features list

>
> Bruce Momjian schrieb:
> > I am continuing on the work, but I am now uncertain I can complete by
> > code freeze in Mid-May, or even by the end of May.
> >
> > I am unsure what we should do:
> >
> > delay code freeze
> > allow Win32 patches during beta
> > delay Win32 for 7.5
> >
> I'm just a "normal" user, but I'd like to throw in my to 2cents as well.
As
> much as I appreciate all the work you are doing, I think that decent Win32
> version will give PostgreSQL big momentum especially over MySQL. I'm
> happily running the Peer-Direct version (although not in a
> business-critical environment) and I would just love to see an "official"
> win32 version.
>
> So from my point of view delaying Win32 for 7.5 would really be a bad
choice.
>
> Cheers
> Thomas
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo(at)postgresql(dot)org

I have gotten some replies so let me chime in now.

Basically, here is how this happens --- you wait for a slow time to do
this major work, and as the deadline draws near, you realize a slow time
isn't coming and you stop doing anything else but the big project. I
bet Patrick is in the same boat.

I am not reading much email anymore, not participating in discussions,
not applying patches, not adding other features --- but I underestimated
the fork/exec issues and now I feel behind. I have phoned Jan and he is
going to help me out, but at this point, I don't know if that will be
enough so I am reporting to the group.

Also, I am reporting problems now, rather than mid-May or June 1.

As far as past delay disasters, the issues were:

o waited until the deadline to extend
o extended deadline in too small increments
o didn't set reasonable new deadlines

So, we have to decide now if we want to delay, and if so, we have to
decide on the new deadline in one month increments (July 1), and we have
to reevalute that on June 1 to determine if July 1 is still
reasonable/desired.

---------------------------------------------------------------------------

Andrew Dunstan wrote:
> Amen.
>
> Since I now have some free time on my hands (through no choice of mine
> ;-( ) I can devote some of it to the W32 work if someone (Bruce?) would
> like to throw some tasks over the fence at me, to help shorten the cycle. (I
> take it from previous discussion what is basically needed is Windows+cygwin,
> right?)
>
> andrew
>
>
> ----- Original Message -----
> From: "Thomas Kellerer" <spam_eater(at)gmx(dot)net>
> To: <pgsql-hackers(at)postgresql(dot)org>
> Sent: Monday, May 05, 2003 5:02 PM
> Subject: Re: [HACKERS] 7.4 features list
>
>
> >
> > Bruce Momjian schrieb:
> > > I am continuing on the work, but I am now uncertain I can complete by
> > > code freeze in Mid-May, or even by the end of May.
> > >
> > > I am unsure what we should do:
> > >
> > > delay code freeze
> > > allow Win32 patches during beta
> > > delay Win32 for 7.5
> > >
> > I'm just a "normal" user, but I'd like to throw in my to 2cents as well.
> As
> > much as I appreciate all the work you are doing, I think that decent Win32
> > version will give PostgreSQL big momentum especially over MySQL. I'm
> > happily running the Peer-Direct version (although not in a
> > business-critical environment) and I would just love to see an "official"
> > win32 version.
> >
> > So from my point of view delaying Win32 for 7.5 would really be a bad
> choice.
> >
> > Cheers
> > Thomas
> >
> >
> > ---------------------------(end of broadcast)---------------------------
> > TIP 1: subscribe and unsubscribe commands go to majordomo(at)postgresql(dot)org
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
> (send "unregister YourEmailAddressHere" to majordomo(at)postgresql(dot)org)
>

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> As far as past delay disasters, the issues were:
> o waited until the deadline to extend
> o extended deadline in too small increments
> o didn't set reasonable new deadlines
> So, we have to decide now if we want to delay, and if so, we have to
> decide on the new deadline in one month increments (July 1), and we have
> to reevalute that on June 1 to determine if July 1 is still
> reasonable/desired.

Do you think there is a substantial chance that you will be done on July
1?

regards, tom lane

Bruce,

I think you are doing the right thing by starting the discussion now.
But there are at least three ways to deal with the potential of missing
the June 1 date. The first way is to move the date a second way is to
cut scope, and a third way is to increase resources. I would like to
focus on that third way. Is there anything any of us out here lurking
on these mail lists can do to help? I know that there are a lot of
people who want to see the win32 port in 7.4. Can the work be broken up
so that others could contribute?

thanks,
--Barry

Bruce Momjian wrote:
> I have gotten some replies so let me chime in now.
>
> Basically, here is how this happens --- you wait for a slow time to do
> this major work, and as the deadline draws near, you realize a slow time
> isn't coming and you stop doing anything else but the big project. I
> bet Patrick is in the same boat.
>
> I am not reading much email anymore, not participating in discussions,
> not applying patches, not adding other features --- but I underestimated
> the fork/exec issues and now I feel behind. I have phoned Jan and he is
> going to help me out, but at this point, I don't know if that will be
> enough so I am reporting to the group.
>
> Also, I am reporting problems now, rather than mid-May or June 1.
>
> As far as past delay disasters, the issues were:
>
> o waited until the deadline to extend
> o extended deadline in too small increments
> o didn't set reasonable new deadlines
>
> So, we have to decide now if we want to delay, and if so, we have to
> decide on the new deadline in one month increments (July 1), and we have
> to reevalute that on June 1 to determine if July 1 is still
> reasonable/desired.
>
> ---------------------------------------------------------------------------
>
> Andrew Dunstan wrote:
>
>>Amen.
>>
>>Since I now have some free time on my hands (through no choice of mine
>>;-( ) I can devote some of it to the W32 work if someone (Bruce?) would
>>like to throw some tasks over the fence at me, to help shorten the cycle. (I
>>take it from previous discussion what is basically needed is Windows+cygwin,
>>right?)
>>
>>andrew
>>
>>
>>----- Original Message -----
>>From: "Thomas Kellerer" <spam_eater(at)gmx(dot)net>
>>To: <pgsql-hackers(at)postgresql(dot)org>
>>Sent: Monday, May 05, 2003 5:02 PM
>>Subject: Re: [HACKERS] 7.4 features list
>>
>>
>>
>>>Bruce Momjian schrieb:
>>>
>>>>I am continuing on the work, but I am now uncertain I can complete by
>>>>code freeze in Mid-May, or even by the end of May.
>>>>
>>>>I am unsure what we should do:
>>>>
>>>>delay code freeze
>>>>allow Win32 patches during beta
>>>>delay Win32 for 7.5
>>>>
>>>
>>>I'm just a "normal" user, but I'd like to throw in my to 2cents as well.
>>
>>As
>>
>>>much as I appreciate all the work you are doing, I think that decent Win32
>>>version will give PostgreSQL big momentum especially over MySQL. I'm
>>>happily running the Peer-Direct version (although not in a
>>>business-critical environment) and I would just love to see an "official"
>>>win32 version.
>>>
>>>So from my point of view delaying Win32 for 7.5 would really be a bad
>>
>>choice.
>>
>>>Cheers
>>>Thomas
>>>
>>>
>>>---------------------------(end of broadcast)---------------------------
>>>TIP 1: subscribe and unsubscribe commands go to majordomo(at)postgresql(dot)org
>>
>>
>>---------------------------(end of broadcast)---------------------------
>>TIP 2: you can get off all lists at once with the unregister command
>> (send "unregister YourEmailAddressHere" to majordomo(at)postgresql(dot)org)
>>
>
>

On Mon, May 05, 2003 at 21:08:37 +0100,
"Nigel J. Andrews" <nandrews(at)investsystems(dot)co(dot)uk> wrote:
>
> I've wondered recently at other mentions of the impending feature freeze and
> wondering why that appears to have been set in stone. I know from comments
> leading to 7.3 that there have been problems in the past with timetable
> slippage but is it really sensible to essentially throw out a major feature
> that will take another 6+ months to be released for the sake of a short delay
> for the coming release?

There is already a lot of good stuff in 7.4. In my opinion what there is
enough to justify a release even without PITR or the WIN32 port.
That isn't to say that it isn't worth waiting a little to save some
testing effort (if say a 7.5 were to come out very shortly after 7.4
with the WIN32 port or PITR), but if it is really going to be a couple
of extra months some consideration should be given to getting the stuff
you already have out to people.

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > As far as past delay disasters, the issues were:
> > o waited until the deadline to extend
> > o extended deadline in too small increments
> > o didn't set reasonable new deadlines
> > So, we have to decide now if we want to delay, and if so, we have to
> > decide on the new deadline in one month increments (July 1), and we have
> > to reevalute that on June 1 to determine if July 1 is still
> > reasonable/desired.
>
> Do you think there is a substantial chance that you will be done on July
> 1?

If it can't be done by July, then I think it isn't worth waiting for. I
am working with Jan now, and he did the fork/exec part of the code, so
let me see what he says tomorrow and I will report back.

Patrick, can you finish PITR by July 1?

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Somewhere in this thread it was mentioned that postgres-r master-slave will
be delayed to 7.5 because of it's lack of funding. How far delayed would
Postgres-r be?

I want both Postgres-r & win32 asap... So.. If PITR, win32 and Postgres-r
will be done soon, but not soon enough to be in 7.4 why not get 7.4 out with
what you have already (start the feature freeze early even) and then get a
7.5 out shortly afterwards? (say 4 months after, instead of 6?)

That would also help with the perception people have of Postgres (The whole
"have we lost momentum" thread). If postgres can be seen getting a number
of releases out quickly, with major new (well tested, stable) features in
each, that will definetly help Postgres' momentum as percieved by the world.

Later
Rob

> On Mon, May 05, 2003 at 21:08:37 +0100,
> "Nigel J. Andrews" <nandrews(at)investsystems(dot)co(dot)uk> wrote:
> >
> > I've wondered recently at other mentions of the impending feature freeze
and
> > wondering why that appears to have been set in stone. I know from
comments
> > leading to 7.3 that there have been problems in the past with timetable
> > slippage but is it really sensible to essentially throw out a major
feature
> > that will take another 6+ months to be released for the sake of a short
delay
> > for the coming release?
>
> There is already a lot of good stuff in 7.4. In my opinion what there is
> enough to justify a release even without PITR or the WIN32 port.
> That isn't to say that it isn't worth waiting a little to save some
> testing effort (if say a 7.5 were to come out very shortly after 7.4
> with the WIN32 port or PITR), but if it is really going to be a couple
> of extra months some consideration should be given to getting the stuff
> you already have out to people.
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster

Bruce Momjian wrote:

> If it can't be done by July, then I think it isn't worth waiting for. I
> am working with Jan now, and he did the fork/exec part of the code, so
> let me see what he says tomorrow and I will report back.
>
> Patrick, can you finish PITR by July 1?

It sounds reasonable but so did May 15. I know what has to be done,
it's just a matter of clearing schedules to get a large block of time.
No vacation until it's done I guess.

Patrick

On Monday 05 May 2003 2:02 pm, Thomas Kellerer wrote:
> Bruce Momjian schrieb:
> > I am continuing on the work, but I am now uncertain I can complete by
> > code freeze in Mid-May, or even by the end of May.
> >
> > I am unsure what we should do:
> >
> > delay code freeze
> > allow Win32 patches during beta
> > delay Win32 for 7.5
>
> I'm just a "normal" user, but I'd like to throw in my to 2cents as well...

As another "normal" user I'll throw in my $.02 as well. PostgreSQL is a
wonderfully stable database server that runs on top of a number of
wonderfully stable OSs. We have several servers running here and see
absolutely no point in converting to Windows as we would have to pay more and
as far as I can tell I switching PostgreSQL to Windows would not give us any
improvements.

We are very interested, however, in PITR and replication as it would save us
a great deal of the effort we currently spend to roll our own redundancy
schemes.

Effort toward PITR, replication and other features moves PostgreSQL forward.
Win32 porting merely spreads limited resources thinner.

So when everyone thinks that there are no significant improvements to be made
or features to be added to PostgreSQL then by all means, start porting. Just
not before.

Cheers,
Steve

Patrick Macdonald wrote:
> Bruce Momjian wrote:
>
> > If it can't be done by July, then I think it isn't worth waiting for. I
> > am working with Jan now, and he did the fork/exec part of the code, so
> > let me see what he says tomorrow and I will report back.
> >
> > Patrick, can you finish PITR by July 1?
>
> It sounds reasonable but so did May 15. I know what has to be done,
> it's just a matter of clearing schedules to get a large block of time.
> No vacation until it's done I guess.

OK, that's the report from Patrick.

Actually, I am more concerned at releasing 7.4 without PITR than I am
releasing without WIn32, because PITR is a feature all platforms can
use, and it is more significant in moving to large installations. We
have Cygwin as a bad workaround for Win32, but we have nothing for PITR.

Let's be frank --- how long have we had the PITR patch? August? How
long have _I_ had the Win32 patches? January? The sad truth is that
these large patches require a deadline to get our plates cleared to work
on them. I know that was true of me, and sounds like Patrick is in the
same boat.

I am talking to Jan regularly and he might drive down from Boston to
Philadelphia to help me for a few days on the Win32 port. I just phoned
Patrick and he says pushing the beta date to July 1 would make it more
likely he will finish, but he is not 100% sure because he has other
responsibilities at Red Hat beside PostgreSQL.

I am not sure what to recommend right now. I would not recommend
pushing beta _past_ July 1. I might have Win32 working for June 1, but
surely July 1, and it is possible PITR might not be ready even on July
1.

Should we wait until July 1 and be sure we have Win32, and maybe PITR?
And if we stick to June 1, will PITR remain dormant like it did after
7.3 and only become active again as we finalized 7.5? Do we need the
pressure of a release to get some of these big features completed?

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> Should we wait until July 1 and be sure we have Win32, and maybe PITR?
> And if we stick to June 1, will PITR remain dormant like it did after
> 7.3 and only become active again as we finalized 7.5? Do we need the
> pressure of a release to get some of these big features completed?

I won't complain too hard if the agreement is to push beta to July 1.
That would make it more likely that we have some frontend support for
the new FE/BE protocol code when we release (right Barry? ;-))

I don't want to see another slip after that, though.

regards, tom lane

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> > Should we wait until July 1 and be sure we have Win32, and maybe PITR?
> > And if we stick to June 1, will PITR remain dormant like it did after
> > 7.3 and only become active again as we finalized 7.5? Do we need the
> > pressure of a release to get some of these big features completed?
>
> I won't complain too hard if the agreement is to push beta to July 1.
> That would make it more likely that we have some frontend support for
> the new FE/BE protocol code when we release (right Barry? ;-))
>
> I don't want to see another slip after that, though.

I am waiting to talk to Jan in a few hours to get his opinion on the
timeframe for Win32. I would like to have a final decision on the beta
date by this Friday.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Tom,

I plan to clear some time very soon to work on moving jdbc to the
version 3 protocol. What is the status of the version 3 protocol
changes, are you done? And yes an extra couple of weeks would ensure
that I get more of the version 3 protocol functionality into the jdbc
driver.

thanks,
--Barry

Tom Lane wrote:
> Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
>
>>Should we wait until July 1 and be sure we have Win32, and maybe PITR?
>>And if we stick to June 1, will PITR remain dormant like it did after
>>7.3 and only become active again as we finalized 7.5? Do we need the
>>pressure of a release to get some of these big features completed?
>
>
> I won't complain too hard if the agreement is to push beta to July 1.
> That would make it more likely that we have some frontend support for
> the new FE/BE protocol code when we release (right Barry? ;-))
>
> I don't want to see another slip after that, though.
>
> regards, tom lane
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>

> I think you are doing the right thing by starting the discussion now.
> But there are at least three ways to deal with the potential of missing
> the June 1 date. The first way is to move the date a second way is to
> cut scope, and a third way is to increase resources. I would like to
> focus on that third way. Is there anything any of us out here lurking
> on these mail lists can do to help? I know that there are a lot of
> people who want to see the win32 port in 7.4. Can the work be broken up
> so that others could contribute?

Well, I can volunteer build testing and runtime testing on our windows
machines here, but that's really about it...

Chris

Barry Lind <blind(at)xythos(dot)com> writes:
> I plan to clear some time very soon to work on moving jdbc to the
> version 3 protocol. What is the status of the version 3 protocol
> changes, are you done?

I have implemented everything I wanted to do at the protocol level
except for fixing the handling of binary values, which is something
I hope to get done in the next few days. It's not too soon to start
working on JDBC's support, if you have free time now. There are lots
of loose ends (for instance, *every* backend error currently returns
the same "XX000" SQLSTATE code) but I do not think there are any
showstoppers for frontend development.

> And yes an extra couple of weeks would ensure that I get more of the
> version 3 protocol functionality into the jdbc driver.

Check. The wind seems to be blowing towards a July 1 beta date.

regards, tom lane

Tom Lane wrote:
> Barry Lind <blind(at)xythos(dot)com> writes:
> > I plan to clear some time very soon to work on moving jdbc to the
> > version 3 protocol. What is the status of the version 3 protocol
> > changes, are you done?
>
> I have implemented everything I wanted to do at the protocol level
> except for fixing the handling of binary values, which is something
> I hope to get done in the next few days. It's not too soon to start
> working on JDBC's support, if you have free time now. There are lots
> of loose ends (for instance, *every* backend error currently returns
> the same "XX000" SQLSTATE code) but I do not think there are any
> showstoppers for frontend development.
>
> > And yes an extra couple of weeks would ensure that I get more of the
> > version 3 protocol functionality into the jdbc driver.
>
> Check. The wind seems to be blowing towards a July 1 beta date.

Tom, I am seeing "IN ... func..." in the server logs for every LOG:
message. Is that planned behavior?

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> Tom, I am seeing "IN ... func..." in the server logs for every LOG:
> message. Is that planned behavior?

Only transient. As I was telling you this afternoon, I'm focusing
on getting the on-the-wire protocol behavior frozen. We need to think
about how much of the new, richer, error information should be logged
in the server log --- and what configuration options we should offer
to control that. But I don't want to think about it quite yet.

regards, tom lane

Tom Lane wrote:
> [...]
> months ago, and they're not done. That gives me no confidence that
> they'll meet the next deadline we set. If September comes and we still
> have no working Win32 port, will you vote to delay again?

I have to agree on this. Things like the Win32 port tend to take longer
and longer. With a release schedule every 3-6 months they'll just get
released whenever they're done. To hold back other improvements because
of them is not right.

Jan

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #

Jan Wieck wrote:
> Tom Lane wrote:
> > [...]
> > months ago, and they're not done. That gives me no confidence that
> > they'll meet the next deadline we set. If September comes and we still
> > have no working Win32 port, will you vote to delay again?
>
> I have to agree on this. Things like the Win32 port tend to take longer
> and longer. With a release schedule every 3-6 months they'll just get
> released whenever they're done. To hold back other improvements because
> of them is not right.

I have been with Jan for two days now and I can report the Win32 status.

Jan is working on getting fork/exec to work in our code. He should be
done next week. He is trying to do it as cleanly as possible.

Here is my Win32 TODO list:

*file path separators
*root directory
*cp, rm
*handle \r in COPY
*rename and unlink atomicity
fork/exec
compatibility defines
spinlock changes
DWORD in help.c
initdb
signals
makefiles and build environment
runtime environment
cancel key handling

UNKNOWN -
handle config file changes that happen before backend starts signals
xcopy vs cp
sync vs commit() SRA

Jan feels that the next step is to decide on a build environment. I
think we had discussed _not_ using the MS project files, but instead
using gmake and perhaps the MingW or Visual C command line compiler.

As far as timeframe, Jan thinks July 1 is going to be a difficult
deadline. However, if most of the code changes are done by then, we can
play with making the Win32 environment and binaries during beta. Of
course, we can probably get the massive changes done by June 1 too.

So, when do we want beta? Seems most want July 1 because Win32 will be
farther along and PITR has a chance of making it. However, none of
those are certain.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

On Fri, 2003-05-09 at 13:28, Bruce Momjian wrote:
> Here is my Win32 TODO list:
> initdb
>
> As far as timeframe, Jan thinks July 1 is going to be a difficult
> deadline. However, if most of the code changes are done by then, we can
> play with making the Win32 environment and binaries during beta. Of
> course, we can probably get the massive changes done by June 1 too.
>

has a decision been reached on the other command line tools like pg_ctl,
pg_config, createlang, etc... ?

> So, when do we want beta? Seems most want July 1 because Win32 will be
> farther along and PITR has a chance of making it. However, none of
> those are certain.
>

ugh... I'd really like to see win32 in the next release if we're going
to do it, but it is a slippery slope.

Robert Treat

On Fri, May 09, 2003 at 15:37:37 -0400,
Robert Treat <xzilla(at)users(dot)sourceforge(dot)net> wrote:
>
> ugh... I'd really like to see win32 in the next release if we're going
> to do it, but it is a slippery slope.

If the beta is July 1, the people working on both PITR and WIN32 will
work hard on those projects trying to make the deadline. This should
result in a lot getting done where as if the pressure was removed
things might happen more slowly. If they just miss they should be close
enough that a quick 7.5 release mainly for WIN32 or PITR might really
be doable. Beta could probably start as soon as 7.4 was released and
a 7.5 release could follow in another month or two.

OK, agreed. Beta will be July 1; feature freeze June 16. You folks
need to keep on top of the Win32 and PITR folks to make sure they keep
working.

---------------------------------------------------------------------------

Bruno Wolff III wrote:
> On Fri, May 09, 2003 at 15:37:37 -0400,
> Robert Treat <xzilla(at)users(dot)sourceforge(dot)net> wrote:
> >
> > ugh... I'd really like to see win32 in the next release if we're going
> > to do it, but it is a slippery slope.
>
> If the beta is July 1, the people working on both PITR and WIN32 will
> work hard on those projects trying to make the deadline. This should
> result in a lot getting done where as if the pressure was removed
> things might happen more slowly. If they just miss they should be close
> enough that a quick 7.5 release mainly for WIN32 or PITR might really
> be doable. Beta could probably start as soon as 7.4 was released and
> a 7.5 release could follow in another month or two.
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 3: if posting/reading through Usenet, please send an appropriate
> subscribe-nomail command to majordomo(at)postgresql(dot)org so that your
> message can get through to the mailing list cleanly
>

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> writes:
> OK, agreed. Beta will be July 1; feature freeze June 16.

Done deal. But no more slips.

regards, tom lane

Bruce Momjian wrote:
> OK, agreed. Beta will be July 1; feature freeze June 16. You folks

One question -- why the feature freeze on June 16th instead of July 1st?
Last September 1st, we did feature freeze and started beta on the same
day, didn't we?

Joe

Joe Conway <mail(at)joeconway(dot)com> writes:
> One question -- why the feature freeze on June 16th instead of July 1st?

We could hardly call it "beta" if there's not a fix-bugs-instead-of-
invent-features period beforehand. I'd be willing to cut a "barely
alpha" release on June 16th ... but to me "beta" means "we have some
reason to think this might work, rather than eat your data". The
"this will likely eat your data" phase is alpha testing ...

regards, tom lane

As a postgresql user, and as someone who develops apps with postgresql,
but does not do development FOR postgresql, I feel compelled to chime in
here and agree.

While I understand that there is a strong opinion that a win32 port will
bring more attention to postgresql, and that this is a good thing, I would
hate to see a release get delayed because of the port.

I have several production systems using postgresql up and running now.
One of these systems has been running for over a year and a half, with no
downtime other than scheduled downtime for upgrades, and one hardware
failure. In spite of this, I still have to fight constant political
battles to keep this application on postgresql. Upper managers who have
never heard of postgresql constantly comment about how maybe it would be
better if we just used sybase or oracle, etc.

Each new feature that gets released for postgresql gives me a little more
ammunition to use in these discussions. Currently, the too biggest areas
of contention are replication and just the "I've never heard of this
'postgres'" factor.

While I know that replication is not coming for 7.4, I think that PITR
would help. At least it would facilitate faster / easier recovery after a
crash... not that I've ever had a crash.

I have to think I'm not the only person in this sort of situation. While
the win32 port may be very beneficial, I would hate to see a situation
where the current installed base had to wait for a feature they needed
because of the win32 port.

Please, this is not meant to disparage the hard work of the win32 porters
at all. I understand that bringing a wider user base into the fold can in
the long term result in more resources to help continue to improve
postgresql. I'm just saying that I would not want to see new features for
the existing ports of postgresql put on hold because of a desire to have
win32 in the next release, if it came down to that situation.

I'd like to thank everyone involved in the postgresql project for creating
such a fantastic tool!

Brian Knox
brian(at)mail(dot)pantalaimon(dot)net

On Thu, 8 May 2003, Jan Wieck wrote:

> Tom Lane wrote:
> > [...]
> > months ago, and they're not done. That gives me no confidence that
> > they'll meet the next deadline we set. If September comes and we still
> > have no working Win32 port, will you vote to delay again?
>
> I have to agree on this. Things like the Win32 port tend to take longer
> and longer. With a release schedule every 3-6 months they'll just get
> released whenever they're done. To hold back other improvements because
> of them is not right.
>
>
> Jan
>
> --
> #======================================================================#
> # It's easier to get forgiveness for being wrong than for being right. #
> # Let's break this rule - forgive me. #
> #================================================== JanWieck(at)Yahoo(dot)com #
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
> (send "unregister YourEmailAddressHere" to majordomo(at)postgresql(dot)org)
>

The delay is for Win32 _and_ PITR, hopefully both for 7.4.

---------------------------------------------------------------------------

Brian wrote:
> As a postgresql user, and as someone who develops apps with postgresql,
> but does not do development FOR postgresql, I feel compelled to chime in
> here and agree.
>
> While I understand that there is a strong opinion that a win32 port will
> bring more attention to postgresql, and that this is a good thing, I would
> hate to see a release get delayed because of the port.
>
> I have several production systems using postgresql up and running now.
> One of these systems has been running for over a year and a half, with no
> downtime other than scheduled downtime for upgrades, and one hardware
> failure. In spite of this, I still have to fight constant political
> battles to keep this application on postgresql. Upper managers who have
> never heard of postgresql constantly comment about how maybe it would be
> better if we just used sybase or oracle, etc.
>
> Each new feature that gets released for postgresql gives me a little more
> ammunition to use in these discussions. Currently, the too biggest areas
> of contention are replication and just the "I've never heard of this
> 'postgres'" factor.
>
> While I know that replication is not coming for 7.4, I think that PITR
> would help. At least it would facilitate faster / easier recovery after a
> crash... not that I've ever had a crash.
>
> I have to think I'm not the only person in this sort of situation. While
> the win32 port may be very beneficial, I would hate to see a situation
> where the current installed base had to wait for a feature they needed
> because of the win32 port.
>
> Please, this is not meant to disparage the hard work of the win32 porters
> at all. I understand that bringing a wider user base into the fold can in
> the long term result in more resources to help continue to improve
> postgresql. I'm just saying that I would not want to see new features for
> the existing ports of postgresql put on hold because of a desire to have
> win32 in the next release, if it came down to that situation.
>
> I'd like to thank everyone involved in the postgresql project for creating
> such a fantastic tool!
>
> Brian Knox
> brian(at)mail(dot)pantalaimon(dot)net
>
>
> On Thu, 8 May 2003, Jan Wieck wrote:
>
> > Tom Lane wrote:
> > > [...]
> > > months ago, and they're not done. That gives me no confidence that
> > > they'll meet the next deadline we set. If September comes and we still
> > > have no working Win32 port, will you vote to delay again?
> >
> > I have to agree on this. Things like the Win32 port tend to take longer
> > and longer. With a release schedule every 3-6 months they'll just get
> > released whenever they're done. To hold back other improvements because
> > of them is not right.
> >
> >
> > Jan
> >
> > --
> > #======================================================================#
> > # It's easier to get forgiveness for being wrong than for being right. #
> > # Let's break this rule - forgive me. #
> > #================================================== JanWieck(at)Yahoo(dot)com #
> >
> >
> > ---------------------------(end of broadcast)---------------------------
> > TIP 2: you can get off all lists at once with the unregister command
> > (send "unregister YourEmailAddressHere" to majordomo(at)postgresql(dot)org)
> >
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
> (send "unregister YourEmailAddressHere" to majordomo(at)postgresql(dot)org)
>

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073