Thread: [HACKERS] PG10 pgindent run
I would like to run pgindent on the head source tree this Wednesday afternoon, UTC. Is that OK for everyone? -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +
On 2017-05-14 09:53:17 -0400, Bruce Momjian wrote: > I would like to run pgindent on the head source tree this Wednesday > afternoon, UTC. Is that OK for everyone? Shouldn't we do that before we branch? And if Thursday still is the intended alpha/beta release date, Wednesday would be too late, no? Greetings, Andres Freund
* Andres Freund (andres@anarazel.de) wrote: > On 2017-05-14 09:53:17 -0400, Bruce Momjian wrote: > > I would like to run pgindent on the head source tree this Wednesday > > afternoon, UTC. Is that OK for everyone? > > Shouldn't we do that before we branch? And if Thursday still is the > intended alpha/beta release date, Wednesday would be too late, no? Releasing alpha/beta is not the same as branching, which I didn't expect us to do for a while yet.. Thanks! Stephen
On 2017-05-14 17:22:16 -0400, Stephen Frost wrote: > * Andres Freund (andres@anarazel.de) wrote: > > On 2017-05-14 09:53:17 -0400, Bruce Momjian wrote: > > > I would like to run pgindent on the head source tree this Wednesday > > > afternoon, UTC. Is that OK for everyone? > > > > Shouldn't we do that before we branch? And if Thursday still is the > > intended alpha/beta release date, Wednesday would be too late, no? > > Releasing alpha/beta is not the same as branching, which I didn't expect > us to do for a while yet.. Well, tagging then. Imo it still should be done before we tag beta1/alpha1. Greetings, Andres Freund
* Andres Freund (andres@anarazel.de) wrote: > On 2017-05-14 17:22:16 -0400, Stephen Frost wrote: > > * Andres Freund (andres@anarazel.de) wrote: > > > On 2017-05-14 09:53:17 -0400, Bruce Momjian wrote: > > > > I would like to run pgindent on the head source tree this Wednesday > > > > afternoon, UTC. Is that OK for everyone? > > > > > > Shouldn't we do that before we branch? And if Thursday still is the > > > intended alpha/beta release date, Wednesday would be too late, no? > > > > Releasing alpha/beta is not the same as branching, which I didn't expect > > us to do for a while yet.. > > Well, tagging then. Imo it still should be done before we tag > beta1/alpha1. For my 2c, at least, I'm not sure I see why that matters all that much..? I'm not against doing the pgindent run today, to be clear, but I don't follow what the harm is with waiting until Wednesday. Thanks! Stephen
Andres Freund <andres@anarazel.de> writes:
> On 2017-05-14 17:22:16 -0400, Stephen Frost wrote:
>> Releasing alpha/beta is not the same as branching, which I didn't expect
>> us to do for a while yet..
> Well, tagging then. Imo it still should be done before we tag
> beta1/alpha1.
Too late, IMO. Yeah, I'd have preferred to pgindent before beta1 too,
just in case pgindent breaks something ... but it's been a long time
since that happened, so realistically, not having done it yet is not
going to meaningfully degrade the relevance of beta testing.
At this point, for Bruce to do it before the wrap, he'd pretty much
have to do it tonight, which we already agreed among the release team
is not enough notice.
Also, I liked the approach Robert took last year where he did a dry
run and then manually fixed places that he saw would get messed up
by pgindent. That's a lot more pleasant than undoing such damage
afterwards. I don't know if Robert plans to do that this year, but if
not, I'm planning to get to it tomorrow or Tuesday. I can't do it
right now because I'm still on higher-priority tasks.
regards, tom lane
Bruce Momjian <bruce@momjian.us> writes:
> I would like to run pgindent on the head source tree this Wednesday
> afternoon, UTC. Is that OK for everyone?
I've been doing some preliminary checking on what pgindent will do,
and I notice that some typedef names are getting misindented because
they are not in the current buildfarm typedef list.
Specifically, we don't seem to have entries for any of the typedefs
associated with the ICU code, eg UChar. This is not terribly
surprising since none of the buildfarm critters contributing typedef
lists are building with --with-icu. It looks like only these
critters are building with ICU:
calliphoridaeculicidaemylodonprionskink
Perhaps one of them could enable typedef collection?
If that can be spun up quickly, I'd recommend holding off the
pgindent run till we get results.
regards, tom lane
On 2017-05-16 18:43:22 -0400, Tom Lane wrote: > Specifically, we don't seem to have entries for any of the typedefs > associated with the ICU code, eg UChar. This is not terribly > surprising since none of the buildfarm critters contributing typedef > lists are building with --with-icu. It looks like only these > critters are building with ICU: > > calliphoridae > culicidae > mylodon > prion > skink > > Perhaps one of them could enable typedef collection? I tried to enable it on calliphoridae, not sure how the min_hours_since logic works, and forced a run on HEAD. Chose calliphoridae because runs are reasonably quick and -DDCOPY_PARSE_PLAN_TREES -DRAW_EXPRESSION_COVERAGE_TEST shouldn't affect typedef results. Shouldn't take too long. No clue if there's some switch that needs to be toggled on the buildfarm side to accept the typedefs, I've never looked at that side of things. - Andres
Andres Freund <andres@anarazel.de> writes:
> No clue if there's some switch that needs to be toggled on the buildfarm
> side to accept the typedefs, I've never looked at that side of things.
AFAIK, not; I think it just takes any typedef reports that aren't too
stale.
regards, tom lane
On 2017-05-16 18:56:57 -0400, Tom Lane wrote: > Andres Freund <andres@anarazel.de> writes: > > No clue if there's some switch that needs to be toggled on the buildfarm > > side to accept the typedefs, I've never looked at that side of things. > > AFAIK, not; I think it just takes any typedef reports that aren't too > stale. Cool. There we go: https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=calliphoridae&dt=2017-05-16%2023:16:53&stg=typedefs
On 05/16/2017 06:54 PM, Andres Freund wrote: > On 2017-05-16 18:43:22 -0400, Tom Lane wrote: >> Specifically, we don't seem to have entries for any of the typedefs >> associated with the ICU code, eg UChar. This is not terribly >> surprising since none of the buildfarm critters contributing typedef >> lists are building with --with-icu. It looks like only these >> critters are building with ICU: >> >> calliphoridae >> culicidae >> mylodon >> prion >> skink >> >> Perhaps one of them could enable typedef collection? > I tried to enable it on calliphoridae, not sure how the min_hours_since > logic works, and forced a run on HEAD. Chose calliphoridae because runs > are reasonably quick and -DDCOPY_PARSE_PLAN_TREES -DRAW_EXPRESSION_COVERAGE_TEST > shouldn't affect typedef results. > > Shouldn't take too long. > > No clue if there's some switch that needs to be toggled on the buildfarm > side to accept the typedefs, I've never looked at that side of things. prion has done a run. UChar is now in the list. cheers andrew -- Andrew Dunstan https://www.2ndQuadrant.com PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Andres Freund <andres@anarazel.de> writes: > There we go: > https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=calliphoridae&dt=2017-05-16%2023:16:53&stg=typedefs Yup, looks good now. Thanks! BTW, comparing the typedef list to what I got a few hours ago, I see that "Function" is now a known type name, along with the UWhatever symbols from ICU. I wonder where that came from? A quick grep suggests that it's not going to mess anything up too badly, but it sure seems like a poor choice for a globally visible typedef name. regards, tom lane
On 05/16/2017 08:39 PM, Tom Lane wrote: > Andres Freund <andres@anarazel.de> writes: >> There we go: >> https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=calliphoridae&dt=2017-05-16%2023:16:53&stg=typedefs > Yup, looks good now. Thanks! > > BTW, comparing the typedef list to what I got a few hours ago, I see > that "Function" is now a known type name, along with the UWhatever > symbols from ICU. I wonder where that came from? A quick grep > suggests that it's not going to mess anything up too badly, but it > sure seems like a poor choice for a globally visible typedef name. > > There is a place in pgindent around line 139 where we filter out typedefs we don't want. You could add it there of you were so inclined. I agree this seems like a remarkably bad choice of name. cheers andrew -- Andrew Dunstan https://www.2ndQuadrant.com PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On Tue, May 16, 2017 at 08:39:36PM -0400, Tom Lane wrote: > Andres Freund <andres@anarazel.de> writes: > > There we go: > > https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=calliphoridae&dt=2017-05-16%2023:16:53&stg=typedefs > > Yup, looks good now. Thanks! > > BTW, comparing the typedef list to what I got a few hours ago, I see > that "Function" is now a known type name, along with the UWhatever > symbols from ICU. I wonder where that came from? A quick grep > suggests that it's not going to mess anything up too badly, but it > sure seems like a poor choice for a globally visible typedef name. OK, I assume we are good to go for Wednesday afternoon, UTC. Thanks for the research. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +
Bruce Momjian <bruce@momjian.us> writes:
> OK, I assume we are good to go for Wednesday afternoon, UTC. Thanks for
> the research.
Yeah, I think we're ready, unless anyone has a large patch they want
to stick in first ...
regards, tom lane
On 2017-05-16 21:02:19 -0400, Tom Lane wrote: > Bruce Momjian <bruce@momjian.us> writes: > > OK, I assume we are good to go for Wednesday afternoon, UTC. Thanks for > > the research. > > Yeah, I think we're ready, unless anyone has a large patch they want > to stick in first ... I've this pending JIT support...
On Tue, May 16, 2017 at 06:06:35PM -0700, Andres Freund wrote: > On 2017-05-16 21:02:19 -0400, Tom Lane wrote: > > Bruce Momjian <bruce@momjian.us> writes: > > > OK, I assume we are good to go for Wednesday afternoon, UTC. Thanks for > > > the research. > > > > Yeah, I think we're ready, unless anyone has a large patch they want > > to stick in first ... > > I've this pending JIT support... That went in April 1. ;-) -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +
On Tue, May 16, 2017 at 09:00:38PM -0400, Bruce Momjian wrote: > On Tue, May 16, 2017 at 08:39:36PM -0400, Tom Lane wrote: > > Andres Freund <andres@anarazel.de> writes: > > > There we go: > > > https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=calliphoridae&dt=2017-05-16%2023:16:53&stg=typedefs > > > > Yup, looks good now. Thanks! > > > > BTW, comparing the typedef list to what I got a few hours ago, I see > > that "Function" is now a known type name, along with the UWhatever > > symbols from ICU. I wonder where that came from? A quick grep > > suggests that it's not going to mess anything up too badly, but it > > sure seems like a poor choice for a globally visible typedef name. > > OK, I assume we are good to go for Wednesday afternoon, UTC. Thanks for > the research. Done. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + As you are, so once was I. As I am, so you will be. + + Ancient Roman grave inscription +