Thread: open items

open items

From
Robert Haas
Date:
Hi,

Just a few reminders about the open items list at
https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items --

- Please don't add issues to this list unless they are the result of
development done during this release cycle. This is not a
general-purpose bug tracker.

- The owner of an item is the person who committed the patch that
caused the problem, because that committer is responsible for cleaning
up the mess. Of course, the patch author is warmly invited to help,
especially if they have aspirations of being a committer some day
themselves. Other help is welcome, too.

- Fixing the stuff on this list is a time-boxed activity. We want to
put out a release on time. If the stuff listed here doesn't get fixed,
the release management team will have to do something about it, like
start yelling at people, or forcing patches to be reverted, which will
be no fun for anyone involved, including but not limited to the
release management team.

A great number of things that were added as open items have already
been resolved, but some of the remaining items have been there for a
while. Here's a quick review of what's on the list as of this moment:

* Incorrect Assert in heap_end/rescan for BHS. Either the description
of this item is inaccurate, or we've been unable to fix an incorrect
assert after more than a month. I interpret
https://www.postgresql.org/message-id/54858BA1-084E-4F7D-B2D1-D15505E512FF%40yesql.se
as a vote in favor of committing some patch by Melanie to fix this.
Either Tomas should commit that patch, or Melanie should commit that
patch, or somebody should say why that patch shouldn't be committed,
or someone should request more help determining whether that patch is
indeed the correct fix, or something. But let's not just sit on this.

* Register ALPN protocol id with IANA. From the mailing list thread,
it is abundantly clear that IANA is in no hurry to finish dealing with
what seems to be a completely pro forma request from our end. I think
we just have to be patient.

* not null constraints break dump/restore. I asked whether all of the
issues had been addressed here and Justin Pryzby opined that the only
thing that was still relevant for this release was a possible test
case change, which I would personally consider a good enough reason to
at least consider calling this done. But it's not clear to me whether
Justin's opinion is the consensus position, or perhaps more
relevantly, whether it's Álvaro's position.

* Temporal PKs allow duplicates with empty ranges. Peter Eisentraut
has started working with Paul Jungwirth on this. Looks good so far.

* Rename sslnegotiation "requiredirect." option to "directonly". I
still think Heikki has implemented the wrong behavior here, and I
don't think this renaming is going to make any difference one way or
the other in how understandable it is. But if we're going to leave the
behavior as-is and do the renaming, then let's get that done.

* Race condition with local injection point detach. Discussion is ongoing.

--
Robert Haas
EDB: http://www.enterprisedb.com



Re: open items

From
Dagfinn Ilmari Mannsåker
Date:
Robert Haas <robertmhaas@gmail.com> writes:

> * Register ALPN protocol id with IANA. From the mailing list thread,
> it is abundantly clear that IANA is in no hurry to finish dealing with
> what seems to be a completely pro forma request from our end. I think
> we just have to be patient.

This appears to have been approved without anyone mentioning it on the
list, and the registry now lists "postgresql" at the bottom:

https://www.iana.org/assignments/tls-extensiontype-values/tls-extensiontype-values.xhtml#alpn-protocol-ids

- ilmari



Re: open items

From
Robert Haas
Date:
On Thu, May 9, 2024 at 3:38 PM Dagfinn Ilmari Mannsåker
<ilmari@ilmari.org> wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
> > * Register ALPN protocol id with IANA. From the mailing list thread,
> > it is abundantly clear that IANA is in no hurry to finish dealing with
> > what seems to be a completely pro forma request from our end. I think
> > we just have to be patient.
>
> This appears to have been approved without anyone mentioning it on the
> list, and the registry now lists "postgresql" at the bottom:
>
> https://www.iana.org/assignments/tls-extensiontype-values/tls-extensiontype-values.xhtml#alpn-protocol-ids

Nice, thanks!

--
Robert Haas
EDB: http://www.enterprisedb.com



Re: open items

From
Michael Paquier
Date:
On Thu, May 09, 2024 at 03:28:13PM -0400, Robert Haas wrote:
> Just a few reminders about the open items list at
> https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items --

Thanks for summarizing the situation.

> * Race condition with local injection point detach. Discussion is ongoing.

I have sent a patch for that yesterday, which I assume is going in the
right direction to close entirely the loop:
https://www.postgresql.org/message-id/Zjx9-2swyNg6E1y1%40paquier.xyz

There is still one point of detail related to the amount of
flexibility we'd want for detachs (concurrent detach happening in
parallel of an automated one in the shmem callback) that I'm not
entirely sure about yet but I've proposed an idea to solve that as
well.  I'm hopeful in getting that wrapped at the beginning of next
week.
--
Michael

Attachment

Re: open items

From
Alvaro Herrera
Date:
On 2024-May-09, Robert Haas wrote:

> * not null constraints break dump/restore. I asked whether all of the
> issues had been addressed here and Justin Pryzby opined that the only
> thing that was still relevant for this release was a possible test
> case change, which I would personally consider a good enough reason to
> at least consider calling this done. But it's not clear to me whether
> Justin's opinion is the consensus position, or perhaps more
> relevantly, whether it's Álvaro's position.

I have fixed the reported issues, so as far as these specific items go,
we could close the reported open item.

However, in doing so I realized that some code is more complex than it
needs to be, and exposes users to ugliness that they don't need to see,
so I posted additional patches.  I intend to get these committed today.

A possible complaint is that the upgrade mechanics which are mostly in
pg_dump with some pieces in pg_upgrade are not very explicitly
documented.  There are already comments in all relevant places, but
perhaps an overall picture is necessary.  I'll see about this, probably
as a long comment somewhere.

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/
"La virtud es el justo medio entre dos defectos" (Aristóteles)



Re: open items

From
Melanie Plageman
Date:
On Thu, May 9, 2024 at 3:28 PM Robert Haas <robertmhaas@gmail.com> wrote:
>
> Just a few reminders about the open items list at
> https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items --
>
> * Incorrect Assert in heap_end/rescan for BHS. Either the description
> of this item is inaccurate, or we've been unable to fix an incorrect
> assert after more than a month. I interpret
> https://www.postgresql.org/message-id/54858BA1-084E-4F7D-B2D1-D15505E512FF%40yesql.se
> as a vote in favor of committing some patch by Melanie to fix this.
> Either Tomas should commit that patch, or Melanie should commit that
> patch, or somebody should say why that patch shouldn't be committed,
> or someone should request more help determining whether that patch is
> indeed the correct fix, or something. But let's not just sit on this.

Sorry, yes, the trivial fix has been done for a while. There is one
outstanding feedback on the patch: an update to one of the comments
suggested by Tomas. I got distracted by trying to repro and fix a bug
from the section "live issues affecting stable branches". I will
update this BHS patch by tonight and commit it once Tomas has a chance
to +1.

Thanks,
Melanie



Re: open items

From
Daniel Gustafsson
Date:
> On 10 May 2024, at 14:48, Melanie Plageman <melanieplageman@gmail.com> wrote:
>
> On Thu, May 9, 2024 at 3:28 PM Robert Haas <robertmhaas@gmail.com> wrote:
>>
>> Just a few reminders about the open items list at
>> https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items --
>>
>> * Incorrect Assert in heap_end/rescan for BHS. Either the description
>> of this item is inaccurate, or we've been unable to fix an incorrect
>> assert after more than a month. I interpret
>> https://www.postgresql.org/message-id/54858BA1-084E-4F7D-B2D1-D15505E512FF%40yesql.se
>> as a vote in favor of committing some patch by Melanie to fix this.

It's indeed a vote for that.

>> Either Tomas should commit that patch, or Melanie should commit that
>> patch, or somebody should say why that patch shouldn't be committed,
>> or someone should request more help determining whether that patch is
>> indeed the correct fix, or something. But let's not just sit on this.
>
> Sorry, yes, the trivial fix has been done for a while. There is one
> outstanding feedback on the patch: an update to one of the comments
> suggested by Tomas. I got distracted by trying to repro and fix a bug
> from the section "live issues affecting stable branches". I will
> update this BHS patch by tonight and commit it once Tomas has a chance
> to +1.

+1

--
Daniel Gustafsson




Re: open items

From
Robert Haas
Date:
On Fri, May 10, 2024 at 8:48 AM Melanie Plageman
<melanieplageman@gmail.com> wrote:
> Sorry, yes, the trivial fix has been done for a while. There is one
> outstanding feedback on the patch: an update to one of the comments
> suggested by Tomas. I got distracted by trying to repro and fix a bug
> from the section "live issues affecting stable branches". I will
> update this BHS patch by tonight and commit it once Tomas has a chance
> to +1.

Great, thanks!

--
Robert Haas
EDB: http://www.enterprisedb.com



Re: open items

From
Robert Haas
Date:
On Fri, May 10, 2024 at 7:14 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> A possible complaint is that the upgrade mechanics which are mostly in
> pg_dump with some pieces in pg_upgrade are not very explicitly
> documented.  There are already comments in all relevant places, but
> perhaps an overall picture is necessary.  I'll see about this, probably
> as a long comment somewhere.

I think that would be really helpful.

--
Robert Haas
EDB: http://www.enterprisedb.com



Re: open items

From
Heikki Linnakangas
Date:
On 09/05/2024 22:39, Robert Haas wrote:
> On Thu, May 9, 2024 at 3:38 PM Dagfinn Ilmari Mannsåker
> <ilmari@ilmari.org> wrote:
>> Robert Haas <robertmhaas@gmail.com> writes:
>>> * Register ALPN protocol id with IANA. From the mailing list thread,
>>> it is abundantly clear that IANA is in no hurry to finish dealing with
>>> what seems to be a completely pro forma request from our end. I think
>>> we just have to be patient.
>>
>> This appears to have been approved without anyone mentioning it on the
>> list, and the registry now lists "postgresql" at the bottom:
>>
>> https://www.iana.org/assignments/tls-extensiontype-values/tls-extensiontype-values.xhtml#alpn-protocol-ids
> 
> Nice, thanks!

Committed the change from "TBD-pgsql" to "postgresql", thanks!

-- 
Heikki Linnakangas
Neon (https://neon.tech)




Re: open items

From
Robert Haas
Date:
Hi,

We are down to three open items, all of which have proposed fixes.
That is great, but we need to keep things moving along, because
according to https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items
we are due to release beta1 on May 23. That means that a release
freeze will be in effect from Saturday, May 18, which is four days
from now. Since committing patches sometimes leads to unexpected
surprises, it would be best if the proposed fixes were put into place
sooner rather than later, to allow time for any adjustments that may
be required.

* Incorrect Assert in heap_end/rescan for BHS
Melanie posted a new patch version 23 hours ago, Michael Paquier
reviewed it 7 hours ago.
See
https://www.postgresql.org/message-id/CAAKRu_a%2B5foybidkmh8FpFAV7iegxetPyPXQ5%3D%2B%2BkqZ%2BZDEUcg%40mail.gmail.com

* Temporal PKs allow duplicates with empty ranges
Peter proposes to revert the feature.
See https://www.postgresql.org/message-id/64c2b2ab-7ce9-475e-ac59-3bfec528bada%40eisentraut.org

* Rename sslnegotiation "requiredirect" option to "directonly"
Latest patch is at
https://www.postgresql.org/message-id/3fdaf4b1-82d1-45bb-8175-f97ff53a1f01%40iki.fi
and, I at least, like it
The basic proposal is to get rid of the idea of having a way to try
both modes (negotiated/direct) and make sslnegotiation just pick one
behavior or the other.

Thanks,

...Robert



Re: open items

From
Michael Paquier
Date:
On Tue, May 14, 2024 at 09:52:35AM -0400, Robert Haas wrote:
> We are down to three open items, all of which have proposed fixes.
> That is great, but we need to keep things moving along, because
> according to https://wiki.postgresql.org/wiki/PostgreSQL_17_Open_Items
> we are due to release beta1 on May 23. That means that a release
> freeze will be in effect from Saturday, May 18, which is four days
> from now. Since committing patches sometimes leads to unexpected
> surprises, it would be best if the proposed fixes were put into place
> sooner rather than later, to allow time for any adjustments that may
> be required.

As of this minute, the open item list is empty @-@.

Thanks all for the various resolutions, updates and commits!
--
Michael

Attachment