Horton
Menu

Re: Adding CI to our tree

Andres Freund <andres@anarazel.de>

View thread
From: Andres Freund <andres@anarazel.de>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Robert Haas <robertmhaas@gmail.com>, Justin Pryzby <pryzby@telsasoft.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Andrew Dunstan <andrew@dunslane.net>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Thomas Munro <thomas.munro@gmail.com>, Melanie Plageman <melanieplageman@gmail.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Daniel Gustafsson <daniel@yesql.se>
Date: 2022-02-13T19:14:56Z
Lists: pgsql-hackers 
Hi,

On 2022-02-13 12:13:17 -0500, Tom Lane wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
> > This is exactly why I'm not a huge fan of having ci stuff in the tree.
> > It supposes that there's one right way to do a build, but in reality,
> > different people want and indeed need to use different options for all
> > kinds of reasons.

Sure. But why is that an argument against "having ci stuff in the tree"?

All it does is to make sure that a certain base-level of testing is easy to
achieve for everyone. I don't like working on windows or mac, but my patches
often have platform dependent bits. Now it's less likely that I need to
manually interact with windows.

I don't think we can (or well should) replace the buildfarm with the CI
stuff. The buildfarm provides extensive and varied coverage for master/release
branches. Which isn't feasible for unmerged development work, including cfbot,
from a resource usage POV alone.


> > That's the whole value of having things like
> > configure and pg_config_manual.h. When we start arguing about whether
> > or ci builds should use -DWRITE_READ_PARSE_PLAN_TREES we're inevitably
> > into the realm where no choice is objectively better,

> Right.  Can we set things up so that it's not too painful to inject
> custom build options into a CI build?

What kind of injection are you thinking about? A patch author can obviously
just add options in .cirrus.yml. That's something possible now, that was not
possible with cfbot applying its own .cirrus.yml

It'd be nice if there were a way to do it more easily for msvc and configure
builds together, right now it'd require modifying those tasks in different
ways. But that's not really a CI question.


I'd like to have things like -fanitize=aligned and
-DWRITE_READ_PARSE_PLAN_TREES on by default for CI, primarily for cfbot's
benefit. Most patch authors won't know about using
-DWRITE_READ_PARSE_PLAN_TREES etc, so they won't even think about enabling
them. We're *really* not doing well on the "timely review" side of things, so
we at least should not waste time on high latency back-forth for easily
automatically detectable things.


> I should think that at the very least one needs to be able to vary the
> configure switches and CPPFLAGS/CFLAGS.

Do you mean as part of a patch tested with cfbot, CI running for pushes to
your own repository, or ...?

Greetings,

Andres Freund

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. ci: enable zstd where available.

  2. ci: compile with -Og where applicable.

  3. ci: include hints how to install OS packages.

  4. ci: fix copy-paste mistake in 16eb8231d1b.

  5. ci: macos: align sysinfo_script to other tasks.

  6. ci: Only use one artifact instruction for logs.

  7. ci: s/CCACHE_SIZE/CCACHE_MAXSIZE/.

  8. pg_basebackup: Skip a few more fsyncs if --no-sync is specified.

  9. TAP tests: check for postmaster.pid anyway when "pg_ctl start" fails.

  10. Don't enable fsync in src/test/recovery/t/008_fsm_truncation.pl.

  11. ci: windows: run initdb with --no-sync.

  12. ci: windows: enable build summary to make it easier to spot warnings / errors.

  13. ci: Add continuous integration for github repositories via cirrus-ci.

  14. Fix TestLib::slurp_file() with offset on windows.