Re: logistics for beta3

On Mon, Jun 28, 2010 at 2:40 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Josh Berkus <josh(at)agliodbs(dot)com> writes:
>> Therefore, I propose that we set a beta3 release date for July 8th.
>> That should give it enough space from the American Holiday.
>
> You mean wrap on Thursday the 8th for release on Monday the 12th?
> That'd be fine with me.  Actual release on the 8th would mean asking
> people to do release prep work when they should be out watching
> fireworks.

AIUI, this is the plan we decided on. So:

- Someone (presumably Bruce) needs to run pgindent. Any reason to
wait any longer on that?
- Someone will need to branch the tree after the wrap and stamp it
9.1devel. Who is doing that?
- We have three (fairly minor) remaining open items that should
ideally be dealt with.

They are:

* bump catalog version for plpython3u change? Use RTLD_GLOBAL? -- I
don't immediately know what the bit about RTLD_GLOBAL is referring to,
and it's not obvious to me that anyone cares about the catversion
issue enough to argue about it. So maybe there's nothing to do here
at all. Anyhow, we should make a decision.
* normalize use of LDFLAGS - I believe Tom is dealing with this. His
proposal seems sensible to me, although as demonstrated just this
week, I am not an expert on the idiosyncracies of linking on different
platforms. This doesn't strike me as a show-stopper for beta3.
* Fix log_temp_files units - I believe the consensus here is to revert
Simon's comment/doco changes and instead make the behavior match the
documentation in both 8.4 and HEAD. I have been assuming Simon would
take care of this since it was his patch, but we're running out of
time. This one is probably good to get fixed before we ship a beta
documenting a behavior we intend to change. If necessary I can pick
this up...

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> - Someone will need to branch the tree after the wrap and stamp it
> 9.1devel. Who is doing that?

Marc generally takes care of making branches.

> * bump catalog version for plpython3u change? Use RTLD_GLOBAL? -- I
> don't immediately know what the bit about RTLD_GLOBAL is referring to,
> and it's not obvious to me that anyone cares about the catversion
> issue enough to argue about it. So maybe there's nothing to do here
> at all. Anyhow, we should make a decision.

Peter was suggesting that if we *stopped* using RTLD_GLOBAL then it
might be possible to use plpython2 and plpython3 concurrently in one
backend. After looking at the archives I'm not convinced that's
workable --- it sounds like not using RTLD_GLOBAL would have the effect
of breaking Python's extension scheme altogether. But at any rate the
ball's in his court to make and test a concrete proposal if he wishes.

> * normalize use of LDFLAGS - I believe Tom is dealing with this.

Yeah, I hope to get that committed today. Any later than today will not
leave enough time for buildfarm testing before the wrap.

regards, tom lane

On Mon, Jul 5, 2010 at 11:01 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> * normalize use of LDFLAGS - I believe Tom is dealing with this.
>
> Yeah, I hope to get that committed today.  Any later than today will not
> leave enough time for buildfarm testing before the wrap.

Hmm. So does that mean we need to get log_temp_files fixed today also?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On Mon, 5 Jul 2010, Robert Haas wrote:

> On Mon, Jun 28, 2010 at 2:40 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Josh Berkus <josh(at)agliodbs(dot)com> writes:
>>> Therefore, I propose that we set a beta3 release date for July 8th.
>>> That should give it enough space from the American Holiday.
>>
>> You mean wrap on Thursday the 8th for release on Monday the 12th?
>> That'd be fine with me.  Actual release on the 8th would mean asking
>> people to do release prep work when they should be out watching
>> fireworks.
>
> AIUI, this is the plan we decided on. So:
>
> - Someone (presumably Bruce) needs to run pgindent. Any reason to
> wait any longer on that?
> - Someone will need to branch the tree after the wrap and stamp it
> 9.1devel. Who is doing that?

Me, after I wrap

----
Marc G. Fournier Hub.Org Hosting Solutions S.A.
scrappy(at)hub(dot)org http://www.hub.org

Yahoo:yscrappy Skype: hub.org ICQ:7615664 MSN:scrappy(at)hub(dot)org

On Mon, Jul 5, 2010 at 11:18 AM, Marc G. Fournier <scrappy(at)hub(dot)org> wrote:
> Me, after I wrap

Cool, thanks.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

Marc G. Fournier wrote:
>>
>> - Someone (presumably Bruce) needs to run pgindent. Any reason to
>> wait any longer on that?
>

The typedefs list on the buildfarm needs to be refreshed. That will take
me some time, since I wasn't aware we were about to do a pg_indent run.

Starting now ...

cheers

andrew

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Mon, Jul 5, 2010 at 11:01 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Yeah, I hope to get that committed today. Any later than today will not
>> leave enough time for buildfarm testing before the wrap.

> Hmm. So does that mean we need to get log_temp_files fixed today also?

No, I'm just concerned about the possibility of needing multiple
buildfarm cycles to shake out platform-specific problems with the
LDFLAGS changes. The log_temp_files business doesn't have any
portability risks AFAICS.

regards, tom lane

Andrew Dunstan wrote:
>
>
> Marc G. Fournier wrote:
>>>
>>> - Someone (presumably Bruce) needs to run pgindent. Any reason to
>>> wait any longer on that?
>>
>
> The typedefs list on the buildfarm needs to be refreshed. That will
> take me some time, since I wasn't aware we were about to do a
> pg_indent run.
>
> Starting now ...
>
>

completed.

cheers

andrew

On Mon, Jul 5, 2010 at 3:14 PM, Andrew Dunstan <andrew(at)dunslane(dot)net> wrote:
> Andrew Dunstan wrote:
>> Marc G. Fournier wrote:
>>>> - Someone (presumably Bruce) needs to run pgindent.  Any reason to
>>>> wait any longer on that?
>>
>> The typedefs list on the buildfarm needs to be refreshed. That will take
>> me some time, since I wasn't aware we were about to do a pg_indent run.
>>
>> Starting now ...
>
> completed.

Cool. So, should we have Bruce go ahead and pgindent now?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company

On Mon, 5 Jul 2010, Robert Haas wrote:

> Cool. So, should we have Bruce go ahead and pgindent now?

Yup, as that will give 3 days before wrap / branch to deal with any fall
out from mit :)

----
Marc G. Fournier Hub.Org Hosting Solutions S.A.
scrappy(at)hub(dot)org http://www.hub.org

Yahoo:yscrappy Skype: hub.org ICQ:7615664 MSN:scrappy(at)hub(dot)org

On mån, 2010-07-05 at 11:01 -0400, Tom Lane wrote:
> Peter was suggesting that if we *stopped* using RTLD_GLOBAL then it
> might be possible to use plpython2 and plpython3 concurrently in one
> backend. After looking at the archives I'm not convinced that's
> workable --- it sounds like not using RTLD_GLOBAL would have the
> effect of breaking Python's extension scheme altogether.

Yeah, plpython regression tests fail of you change RTLD_GLOBAL to
RTLD_LOCAL.

I will make a note in the documentation that using plpython2 and 3
together doesn't work.

Robert Haas wrote:
> On Mon, Jul 5, 2010 at 3:14 PM, Andrew Dunstan <andrew(at)dunslane(dot)net> wrote:
> > Andrew Dunstan wrote:
> >> Marc G. Fournier wrote:
> >>>> - Someone (presumably Bruce) needs to run pgindent. ?Any reason to
> >>>> wait any longer on that?
> >>
> >> The typedefs list on the buildfarm needs to be refreshed. That will take
> >> me some time, since I wasn't aware we were about to do a pg_indent run.
> >>
> >> Starting now ...
> >
> > completed.
>
> Cool. So, should we have Bruce go ahead and pgindent now?

Done and committed.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ None of us is going to be here forever. +