Обсуждение: Re: [COMMITTERS] pgsql: configure tag'd 8.3.0 and built witih autoconf 2.59

Andrew Dunstan <andrew@dunslane.net> writes:
> Bruce Momjian wrote:
>> Andrew Dunstan wrote:
>>> I don't see the branch point for REL8_3_STABLE - has that been done? I 
>>> thought it would happen at the same time as we tagged the release.
>> 
>> No, we will branch later.

> We need this to be created before buildfarm members can start testing 8.3.

Nonsense.  At the moment, HEAD is 8.3.

> I still don't see why we wait to create the branch after tagging the 
> release. There doesn't seem to be any point.

To avoid double-patching effort.  I think we'll branch fairly shortly,
like in a week or so, but right now it'd mostly just create make-work
for committers.
        regards, tom lane

Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
>   
>> Bruce Momjian wrote:
>>     
>>> Andrew Dunstan wrote:
>>>       
>>>> I don't see the branch point for REL8_3_STABLE - has that been done? I 
>>>> thought it would happen at the same time as we tagged the release.
>>>>         
>>> No, we will branch later.
>>>       
>
>   
>> We need this to be created before buildfarm members can start testing 8.3.
>>     
>
> Nonsense.  At the moment, HEAD is 8.3.
>   

True, but it won't show up under that heading, and buildfarm members 
won't be able to configure specific 8.3 testing until we branch.

>   
>> I still don't see why we wait to create the branch after tagging the 
>> release. There doesn't seem to be any point.
>>     
>
> To avoid double-patching effort.  I think we'll branch fairly shortly,
> like in a week or so, but right now it'd mostly just create make-work
> for committers.
>
>
>   

Was that a big problem last release?

cheers

andrew

Andrew Dunstan <andrew@dunslane.net> writes:
> Tom Lane wrote:
>> To avoid double-patching effort.  I think we'll branch fairly shortly,
>> like in a week or so, but right now it'd mostly just create make-work
>> for committers.

> Was that a big problem last release?

Well, basically this happens at core's discretion, and we all feel that
waiting a bit more will minimize work.  If there were a lot of people
chomping at the bit to start committing 8.4-only stuff, maybe we'd
decide differently.  But right now what's on my radar screen is still
8.3 bugs, eg the open patch for bug #3921, and we're expecting a few
more new reports as soon as 8.3.0 spreads.

As best I recall, the immediate branch after 8.2 was the exception not
the rule --- we've usually waited longer than that.
        regards, tom lane

Tom Lane wrote:
> As best I recall, the immediate branch after 8.2 was the exception not
> the rule --- we've usually waited longer than that.

8.2, 8.1, and 8.0 were branched off the x.y.0 release tag.  7.4 was branched 
at rc1, 7.3 was branched at beta4, 7.2 was branched at final release, 7.1 was 
branched at 7.1.1.  So no. :-)

Personally, I don't agree with following this endless release cycle with an 
additional indefinite waiting period.

-- 
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut wrote:
> Tom Lane wrote:
>   
>> As best I recall, the immediate branch after 8.2 was the exception not
>> the rule --- we've usually waited longer than that.
>>     
>
> 8.2, 8.1, and 8.0 were branched off the x.y.0 release tag.  7.4 was branched 
> at rc1, 7.3 was branched at beta4, 7.2 was branched at final release, 7.1 was 
> branched at 7.1.1.  So no. :-)
>
> Personally, I don't agree with following this endless release cycle with an 
> additional indefinite waiting period.
>   

Yeah, that accords with my recollection. Also, it also would have been 
nice if there had been some information about what was going on.

I don't really buy the double patching argument. Back patching becomes 
more difficult when there has been significant code drit, but we surely 
don't expect that much drift in the next week or two. Back patching when 
there has been no code drift is pretty simple.

cheers

andrew

Andrew Dunstan <andrew@dunslane.net> writes:
> I don't really buy the double patching argument. Back patching becomes 
> more difficult when there has been significant code drit, but we surely 
> don't expect that much drift in the next week or two. Back patching when 
> there has been no code drift is pretty simple.

Well, it's not hard, but it is tedious.  Bruce and I, who are the people
most likely to bear the brunt of such tedium, both voted to wait a week
or so before branching.  Peter did not bother to vote.
        regards, tom lane

On Mon, Feb 04, 2008 at 08:36:47PM -0500, Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
> > I don't really buy the double patching argument. Back patching becomes 
> > more difficult when there has been significant code drit, but we surely 
> > don't expect that much drift in the next week or two. Back patching when 
> > there has been no code drift is pretty simple.
> 
> Well, it's not hard, but it is tedious.  Bruce and I, who are the people
> most likely to bear the brunt of such tedium, both voted to wait a week
> or so before branching.  Peter did not bother to vote.

I assume this vote was taken out on -core? I don't mind -core deciding on
this, not at all, but I would appreciate it if you would post the result of
the vote on -hackers. It makes a lot of difference with an open-ended
"we'll branch sometmie later" and a "we talked about it, and we decided
we'll branch in one to two weeks unless something unusual comes up".

If you alraedy did this and I missed it in the mail-flood around fixing all
the presskits, I apologize in advance ;-)

//Magnus

On Feb 5, 2008 9:00 AM, Magnus Hagander <magnus@hagander.net> wrote:

> I assume this vote was taken out on -core? I don't mind -core deciding on
> this, not at all, but I would appreciate it if you would post the result of
> the vote on -hackers.

It wasn't a 'vote' in the formal sense. It was just a discussion with
people airing their opinion.

/D

On Tue, Feb 05, 2008 at 10:57:16AM +0000, Dave Page wrote:
> On Feb 5, 2008 9:00 AM, Magnus Hagander <magnus@hagander.net> wrote:
> 
> > I assume this vote was taken out on -core? I don't mind -core deciding on
> > this, not at all, but I would appreciate it if you would post the result of
> > the vote on -hackers.
> 
> It wasn't a 'vote' in the formal sense. It was just a discussion with
> people airing their opinion.

*how* you came to the decision isn't really what I care about in this
case..

//Magnus

Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
>   
>> I don't really buy the double patching argument. Back patching becomes 
>> more difficult when there has been significant code drit, but we surely 
>> don't expect that much drift in the next week or two. Back patching when 
>> there has been no code drift is pretty simple.
>>     
>
> Well, it's not hard, but it is tedious.  Bruce and I, who are the people
> most likely to bear the brunt of such tedium, both voted to wait a week
> or so before branching.  Peter did not bother to vote.
>
>   

I suspect that you made this decision thinking that it didn't affect 
anybody else much. But it does affect buildfarm members. The buildfarm 
requires manual adjustment for each new branch to be built. Up to now 
(as Peter showed) owners have been able to say "Oh, there's a new 
release. I'll start building the new branch". With the branch delayed 
they will have to say "Oh, there's a new release. I wonder when they 
will branch so I can start building the new branch." I suspect there are 
some buildfarm owners who don't read -hackers religiously, and who will 
be somewhat in the dark.

This probably wasn't on the core team's horizon - IIRC Dave is the only 
member of core who runs a buildfarm member.

cheers

andrew

On Feb 5, 2008 11:50 AM, Andrew Dunstan <andrew@dunslane.net> wrote:
>
> This probably wasn't on the core team's horizon - IIRC Dave is the only
> member of core who runs a buildfarm member.

To be honest the zoo beside me didn't even cross my mind when that
thread happened. I didn't pay much attention as it's doesn't affect me
much - plus I've spent the last 2 months double patching pgAdmin since
the EDB QA team started finding all sorts of obscure buglets, so I can
see where Tom's coming from.

/D

"Andrew Dunstan" <andrew@dunslane.net> writes:

> Tom Lane wrote:
>
> With the branch delayed they will have to say "Oh, there's a new
> release. I wonder when they will branch so I can start building the new
> branch." 

Doesn't that just mean they should be sure to announce the branch loudly when
it happens? I can't really see them failing to do that anyways.

Whether the branch or not is just an procedural detail that only committers
need to be concerned with. I think this is all projected anguish over
something related. Tom says "if there were people chomping at the bit to
commit" but that just raises the question: why aren't there people chomping at
said bit? There are certainly tons of patches queued up in Bruce's "held for
8.4" queue.

Is it that just that we have queued up 8.3.x patches occupying committers
attention? Or is it that reviewers and committers are exhausted from 8.3 and
not quite ready yet to tackle new patches?

Either way we need to find a solution that doesn't overload reviewers and
committers and makes continual progress against pending patches. If we don't
keep the patch queue short now we're only going to have a repeat of the
last-minute rush at feature-freeze time again.

--  Gregory Stark EnterpriseDB          http://www.enterprisedb.com Ask me about EnterpriseDB's Slony Replication
support!

Gregory Stark wrote:
> "Andrew Dunstan" <andrew@dunslane.net> writes:
>
>   
>> Tom Lane wrote:
>>
>> With the branch delayed they will have to say "Oh, there's a new
>> release. I wonder when they will branch so I can start building the new
>> branch." 
>>     
>
>
>   

No, I wrote that, not Tom. Your snipping went slightly astray.

cheers

andrew

"Andrew Dunstan" <andrew@dunslane.net> writes:

> Gregory Stark wrote:
>> "Andrew Dunstan" <andrew@dunslane.net> writes:
>>
>>> Tom Lane wrote:
>>>
>>> With the branch delayed they will have to say "Oh, there's a new
>>> release. I wonder when they will branch so I can start building the new
>>> branch."     
>
> No, I wrote that, not Tom. Your snipping went slightly astray.

Er, yeah. Whatever Tom wrote would have had another level of >s. 
Obviously I should have snipped one more line.

--  Gregory Stark EnterpriseDB          http://www.enterprisedb.com Ask me about EnterpriseDB's Slony Replication
support!

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



- --On Tuesday, February 05, 2008 10:00:29 +0100 Magnus Hagander 
<magnus@hagander.net> wrote:

> On Mon, Feb 04, 2008 at 08:36:47PM -0500, Tom Lane wrote:
>> Andrew Dunstan <andrew@dunslane.net> writes:
>> > I don't really buy the double patching argument. Back patching becomes
>> > more difficult when there has been significant code drit, but we surely
>> > don't expect that much drift in the next week or two. Back patching when
>> > there has been no code drift is pretty simple.
>>
>> Well, it's not hard, but it is tedious.  Bruce and I, who are the people
>> most likely to bear the brunt of such tedium, both voted to wait a week
>> or so before branching.  Peter did not bother to vote.
>
> I assume this vote was taken out on -core? I don't mind -core deciding on
> this, not at all, but I would appreciate it if you would post the result of
> the vote on -hackers. It makes a lot of difference with an open-ended
> "we'll branch sometmie later" and a "we talked about it, and we decided
> we'll branch in one to two weeks unless something unusual comes up".

Actually, "branch in one to two weeks" has been the status quo almost since day 
one ... not that I'm against "branch on release", I'm only saying that we've 
followed this same procedure on branching since ... forever.

- ----
Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
Email . scrappy@hub.org                              MSN . scrappy@hub.org
Yahoo . yscrappy               Skype: hub.org        ICQ . 7615664
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.4 (FreeBSD)

iD8DBQFHqTeE4QvfyHIvDvMRAlisAKCxactS3Xp6V9/PbOOn11vhPioQaACgm+Ck
psuY9S9odAYdj91T5/QlYDc=
=CdzR
-----END PGP SIGNATURE-----

Marc G. Fournier wrote:
> Actually, "branch in one to two weeks" has been the status quo almost since
> day one ... not that I'm against "branch on release", I'm only saying that
> we've followed this same procedure on branching since ... forever.

That is incorrect.  See earlier in this thread.

-- 
Peter Eisentraut
http://developer.postgresql.org/~petere/