Re: 8.3 Development Cycle

Following the recent discussion on this list and another on pgsql-core,
we have decided that we would like to aim to meet the following schedule
for the release of PostgreSQL 8.3:

April 1st 2007 - Feature freeze

May 1st 2007 - Beta 1 release

June 1st 2007 - Release

This will obviously be a short development cycle which will allow us to
get some of the features that just missed 8.2 out of the door, as well
as giving us the opportunity to try releasing before the summer (for
those in the northern hemisphere) rather than after.

We are also aware that this is a tight timetable, however given the
shorter development cycle we feel it is an achievable goal.

If anyone has any serious objections, please shout now!

Regards, Dave

--
Dave Page
PostgreSQL Core Team

Dave Page wrote:
> Following the recent discussion on this list and another on pgsql-core,
> we have decided that we would like to aim to meet the following schedule
> for the release of PostgreSQL 8.3:
>
> April 1st 2007 - Feature freeze
>
> May 1st 2007 - Beta 1 release
>
> June 1st 2007 - Release
>
> This will obviously be a short development cycle which will allow us to
> get some of the features that just missed 8.2 out of the door, as well
> as giving us the opportunity to try releasing before the summer (for
> those in the northern hemisphere) rather than after.
>
> We are also aware that this is a tight timetable, however given the
> shorter development cycle we feel it is an achievable goal.
>
> If anyone has any serious objections, please shout now!

Sounds fine, but announcing this now is almost certain to reduce the
number of people migrating to 8.2. I am not saying we shouldn't
announce it now, but it is something I wanted to mention.

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

+ If your life is a hard drive, Christ can be your backup. +

> -----Original Message-----
> From: Bruce Momjian [mailto:bruce(at)momjian(dot)us]
> Sent: 22 September 2006 15:26
> To: Dave Page
> Cc: pgsql-hackers(at)postgresql(dot)org
> Subject: Re: [HACKERS] 8.3 Development Cycle
>
> Sounds fine, but announcing this now is almost certain to reduce the
> number of people migrating to 8.2. I am not saying we shouldn't
> announce it now, but it is something I wanted to mention.

Err right - if you had said this yesterday when we discussed the idea I
could have mentioned it in my message. Right around the time you said
"Go for it" would have been good :-)

Regards, Dave.

Dave Page wrote:
>
>
> > -----Original Message-----
> > From: Bruce Momjian [mailto:bruce(at)momjian(dot)us]
> > Sent: 22 September 2006 15:26
> > To: Dave Page
> > Cc: pgsql-hackers(at)postgresql(dot)org
> > Subject: Re: [HACKERS] 8.3 Development Cycle
> >
> > Sounds fine, but announcing this now is almost certain to reduce the
> > number of people migrating to 8.2. I am not saying we shouldn't
> > announce it now, but it is something I wanted to mention.
>
> Err right - if you had said this yesterday when we discussed the idea I
> could have mentioned it in my message. Right around the time you said
> "Go for it" would have been good :-)

I didn't think of it until today. Sorry. Personally, I don't like
manipulating people by withholding information, so I still think we are
doing the right thing. I am just saying it might have that effect.

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

+ If your life is a hard drive, Christ can be your backup. +

Bruce Momjian wrote:
> Dave Page wrote:
>
>>
>>
>>
>>> -----Original Message-----
>>> From: Bruce Momjian [mailto:bruce(at)momjian(dot)us]
>>> Sent: 22 September 2006 15:26
>>> To: Dave Page
>>> Cc: pgsql-hackers(at)postgresql(dot)org
>>> Subject: Re: [HACKERS] 8.3 Development Cycle
>>>
>>> Sounds fine, but announcing this now is almost certain to reduce the
>>> number of people migrating to 8.2. I am not saying we shouldn't
>>> announce it now, but it is something I wanted to mention.
>>>
>> Err right - if you had said this yesterday when we discussed the idea I
>> could have mentioned it in my message. Right around the time you said
>> "Go for it" would have been good :-)
>>
>
> I didn't think of it until today. Sorry. Personally, I don't like
> manipulating people by withholding information, so I still think we are
> doing the right thing. I am just saying it might have that effect.
>
>

They will upgrade if they want the new features, as usual. I had dinner
with 3 significant users yesterday and mentioned to them the new
facility to add or drop inheritance on tables, and their reaction was
along the lines of "When can I have it? I want it now!" They won't be
deterred by the fact that the next dev cycle will be a bit shorter.

cheers

andrew

> -----Original Message-----
> From: Bruce Momjian [mailto:bruce(at)momjian(dot)us]
> Sent: 22 September 2006 15:35
> To: Dave Page
> Cc: pgsql-hackers(at)postgresql(dot)org
> Subject: Re: [HACKERS] 8.3 Development Cycle
>
> > Err right - if you had said this yesterday when we
> discussed the idea I
> > could have mentioned it in my message. Right around the
> time you said
> > "Go for it" would have been good :-)
>
> I didn't think of it until today. Sorry.

NP - I wrote that firmly tongue in cheek, hence the smiley.

> Personally, I don't like
> manipulating people by withholding information, so I still
> think we are
> doing the right thing.

Yes, agreed. Hopefully it will help people to plan their time given as
much advance notice of cycle dates as possible.

> I am just saying it might have that effect.

For some people I'm sure it will. Others (myself included) will want to
move to 8.2 as soon as possible.

Regards, dave.

Dave Page wrote:
>
>
>> -----Original Message-----
>> From: Bruce Momjian [mailto:bruce(at)momjian(dot)us]
>> Sent: 22 September 2006 15:26
>> To: Dave Page
>> Cc: pgsql-hackers(at)postgresql(dot)org
>> Subject: Re: [HACKERS] 8.3 Development Cycle
>>
>> Sounds fine, but announcing this now is almost certain to reduce the
>> number of people migrating to 8.2. I am not saying we shouldn't
>> announce it now, but it is something I wanted to mention.
>
> Err right - if you had said this yesterday when we discussed the idea I
> could have mentioned it in my message. Right around the time you said
> "Go for it" would have been good :-)

This will likely stop people from migrating to 8.2, but so what? It
isn't going to stop new users and existing users in real production
setting will likely wait for 8.3 anyway.

Joshua D. Drake

>

--

=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240
Providing the most comprehensive PostgreSQL solutions since 1997
http://www.commandprompt.com/

Bruce, Dave,

> This will likely stop people from migrating to 8.2, but so what? It
> isn't going to stop new users and existing users in real production
> setting will likely wait for 8.3 anyway.

And at this point most production users are only upgrading every 2-3
releases anyway (something which will get worse with time). Heck, I
have former clients who are still running 7.2. Why would they upgrade?
It's never been down, and it's not exposed to untrusted users.

--Josh Berkus

Tom,

> I'm obviously thinking of enums which was ready (for review at least) a
> few weeks ago, but has probably bitrotted slightly since then given the
> number of patches that have landed in the tree. I intended to brush it
> up as soon as the 8.3 tree was open and resubmit it. Will that be a
> waste of time?

Ooops. Are you sure these weren't committed?

--Josh

Tom Dunstan <pgsql(at)tomd(dot)cc> writes:
> Joshua's original mail suggested that only certain features would go in.
> Is that still on the cards, or will other features be considered if
> they're ready?

You'll note that Dave's mail said no such thing.

There has been some talk of trying to agree on a roadmap for 8.3 and
coordinate development efforts accordingly --- but that does not
constitute an agreement to reject work not in the roadmap, just some
coordination among those people who wish to coordinate.

> I'm obviously thinking of enums which was ready (for review at least) a
> few weeks ago, but has probably bitrotted slightly since then given the
> number of patches that have landed in the tree. I intended to brush it
> up as soon as the 8.3 tree was open and resubmit it.

Please do.

regards, tom lane

Tom Dunstan wrote:
> Dave Page wrote:
>> This will obviously be a short development cycle which will allow us to
>> get some of the features that just missed 8.2 out of the door, as well
>> as giving us the opportunity to try releasing before the summer (for
>> those in the northern hemisphere) rather than after.
>
> Joshua's original mail suggested that only certain features would go in.
> Is that still on the cards, or will other features be considered if
> they're ready?
>
> I'm obviously thinking of enums which was ready (for review at least) a
> few weeks ago, but has probably bitrotted slightly since then given the
> number of patches that have landed in the tree. I intended to brush it
> up as soon as the 8.3 tree was open and resubmit it. Will that be a
> waste of time?
>

AFAIK, we aren't doing anything like that, and it would be quite unfair
if we did.

cheers

andrew

Josh Berkus <josh(at)agliodbs(dot)com> writes:
> Bruce, Dave,
>> This will likely stop people from migrating to 8.2, but so what? It
>> isn't going to stop new users and existing users in real production
>> setting will likely wait for 8.3 anyway.

> And at this point most production users are only upgrading every 2-3
> releases anyway (something which will get worse with time).

The other side of that coin is that with a short devel cycle, 8.3 is
not necessarily going to look like a must-have upgrade to many people
either. If I were a DBA looking at the current plans, and I didn't have
a desperate need for bitmap indexes (a feature with a still very unclear
use-case footprint ...), I'd probably figure that updating to 8.2 soon
is a more rewarding strategy than waiting for 8.3. Known benefits now
versus unknown benefits later is a pretty easy call.

In the end, any one user is going to find particular updates compelling
or not based on specific features they need for their specific
application. We can't any longer expect that everyone's going to adopt
every release immediately ... indeed, that's why we're still supporting
back release branches.

regards, tom lane

On Fri, Sep 22, 2006 at 02:16:53PM +0100, Dave Page wrote:
> Following the recent discussion on this list and another on
> pgsql-core, we have decided that we would like to aim to meet the
> following schedule for the release of PostgreSQL 8.3:
>
> April 1st 2007 - Feature freeze
^^^^^^^^^

We should probably move this forward or back one day.

Cheers,
D
--
David Fetter <david(at)fetter(dot)org> http://fetter.org/
phone: +1 415 235 3778 AIM: dfetter666
Skype: davidfetter

Remember to vote!

Josh Berkus wrote:
> Tom,
>
>> I'm obviously thinking of enums which was ready (for review at least) a
>> few weeks ago, but has probably bitrotted slightly since then given the
>> number of patches that have landed in the tree. I intended to brush it
>> up as soon as the 8.3 tree was open and resubmit it. Will that be a
>> waste of time?
>
> Ooops. Are you sure these weren't committed?
>
>
It's not committed - it was submitted long after feature freeze (in
fact, coding didn't begin until after freeze). But it's fairly liable to
bitrot, since it touches the catalog in a significant way.

cheers

andrew

Tom,

> Pretty sure. :) Why the oops? They haven't been mentioned in some PR
> material or something have they?

No, I'd just been confused and thought the patch was submitted before
feature freeze.

--
--Josh

Josh Berkus
PostgreSQL @ Sun
San Francisco

Dave Page wrote:
> This will obviously be a short development cycle which will allow us to
> get some of the features that just missed 8.2 out of the door, as well
> as giving us the opportunity to try releasing before the summer (for
> those in the northern hemisphere) rather than after.

Joshua's original mail suggested that only certain features would go in.
Is that still on the cards, or will other features be considered if
they're ready?

I'm obviously thinking of enums which was ready (for review at least) a
few weeks ago, but has probably bitrotted slightly since then given the
number of patches that have landed in the tree. I intended to brush it
up as soon as the 8.3 tree was open and resubmit it. Will that be a
waste of time?

Thanks

Tom

On Friday 22 September 2006 12:40, Tom Lane wrote:
> Tom Dunstan <pgsql(at)tomd(dot)cc> writes:
> > Joshua's original mail suggested that only certain features would go in.
> > Is that still on the cards, or will other features be considered if
> > they're ready?
>
> You'll note that Dave's mail said no such thing.

I'd like to see some type of statement from core what level of changes thier
willing to allow in the core for this short release (system catalog changes
seem like a must, and looks like on disk format changes is ok too?) Of
course my angle is making the upgrade from 8.2->8.3 as painless as
possible... if we can avoid a dump/reload cycle then people are less likely
to have to choose between 8.2 and 8.3, which would make everyone happy I
imagine.

--
Robert Treat
Build A Brighter LAMP :: Linux Apache {middleware} PostgreSQL

Robert Treat wrote:
> On Friday 22 September 2006 12:40, Tom Lane wrote:
> > Tom Dunstan <pgsql(at)tomd(dot)cc> writes:
> > > Joshua's original mail suggested that only certain features would go in.
> > > Is that still on the cards, or will other features be considered if
> > > they're ready?
> >
> > You'll note that Dave's mail said no such thing.
>
> I'd like to see some type of statement from core what level of changes thier
> willing to allow in the core for this short release (system catalog changes
> seem like a must, and looks like on disk format changes is ok too?) Of
> course my angle is making the upgrade from 8.2->8.3 as painless as
> possible... if we can avoid a dump/reload cycle then people are less likely
> to have to choose between 8.2 and 8.3, which would make everyone happy I
> imagine.

Agreed, but my guess is that we are going to introduce shorter varlena
headers for 8.3. It will hard to reject an optimization like that, and
that will probably change the disk format for most columns.

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

+ If your life is a hard drive, Christ can be your backup. +

Josh Berkus wrote:
>> I'm obviously thinking of enums which was ready (for review at least) a
>> few weeks ago, but has probably bitrotted slightly since then given the
>> number of patches that have landed in the tree. I intended to brush it
>> up as soon as the 8.3 tree was open and resubmit it. Will that be a
>> waste of time?
>
> Ooops. Are you sure these weren't committed?

Pretty sure. :) Why the oops? They haven't been mentioned in some PR
material or something have they?

Cheers

Tom

Bruce Momjian <bruce(at)momjian(dot)us> writes:
> Agreed, but my guess is that we are going to introduce shorter varlena
> headers for 8.3. It will hard to reject an optimization like that, and
> that will probably change the disk format for most columns.

Well, several of the proposals that have been made would not cause
existing disk images to become broken --- in particular, the idea of
introducing separate "short" datatypes without touching the existing
ones would have that merit. So we might want to factor that point into
our choices about what to do.

This is all pretty pointless unless someone actually writes a pg_upgrade
tool, but maybe we'll see one for 8.3.

regards, tom lane

Tom Lane wrote:
> Tom Dunstan <pgsql(at)tomd(dot)cc> writes:
>> Joshua's original mail suggested that only certain features would go in.
>> Is that still on the cards, or will other features be considered if
>> they're ready?
>
> You'll note that Dave's mail said no such thing.

No, but it did explicitly mention features that just missed 8.2, so I
just wanted some clarification, which you and Dave have now provided.
Thanks.

Tom

On Fri, 2006-09-22 at 14:16 +0100, Dave Page wrote:
> Following the recent discussion on this list and another on pgsql-core,
> we have decided that we would like to aim to meet the following schedule
> for the release of PostgreSQL 8.3:
>
> April 1st 2007 - Feature freeze
>
> May 1st 2007 - Beta 1 release
>
> June 1st 2007 - Release
>
> This will obviously be a short development cycle which will allow us to
> get some of the features that just missed 8.2 out of the door, as well
> as giving us the opportunity to try releasing before the summer (for
> those in the northern hemisphere) rather than after.
>
> We are also aware that this is a tight timetable, however given the
> shorter development cycle we feel it is an achievable goal.

Thanks for that, Dave and Core.

I'm very happy to have clearly stated dates. That means we can all plan
what we'll be able to achieve in that time, which is important when some
of the largest or most complex features are being considered.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com