Horton
Menu

Re: run pgindent on a regular basis / scripted manner

Noah Misch <noah@leadboat.com>

View thread
From: Noah Misch <noah@leadboat.com>
To: Magnus Hagander <magnus@hagander.net>
Cc: Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, Bruce Momjian <bruce@momjian.us>, Jelte Fennema <postgres@jeltef.nl>, Peter Geoghegan <pg@bowt.ie>, Alvaro Herrera <alvherre@2ndquadrant.com>, Stephen Frost <sfrost@snowman.net>, Jesse Zhang <sbjesse@gmail.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2023-04-22T19:59:06Z
Lists: pgsql-hackers 
(Given that another commentator is "absolutely against" a hook, this message
is mostly for readers considering this for other projects.)

On Sat, Apr 22, 2023 at 03:23:59PM +0200, Magnus Hagander wrote:
> On Tue, Feb 7, 2023 at 5:43 AM Noah Misch <noah@leadboat.com> wrote:
> > On Mon, Feb 06, 2023 at 06:17:02PM +0100, Peter Eisentraut wrote:
> > > Also, pgindent takes tens of seconds to run, so hooking that into the git
> > > push process would slow this down quite a bit.
> >
> > The pre-receive hook would do a full pgindent when you change typedefs.list.
> > Otherwise, it would reindent only the files being changed.  The average push
> > need not take tens of seconds.
> 
> It would probably ont be tens of seconds, but it would be slow. It
> would need to do a clean git checkout into an isolated environment and
> spawn in there, and just that takes time.

That would be slow, but I wouldn't do it that way.  I'd make "pg_bsd_ident
--pre-receive --show-diff" that, instead of reading from the filesystem, gets
the bytes to check from the equivalent of this Perl-like pseudocode:

while (<>) {
    my($old_hash, $new_hash, $ref) = split;
    foreach my $filename (split /\n/, `git diff --name-only $old_hash..$new_hash`) {
        $file_content = `git show $new_hash $filename`;
    }
}

I just ran pgindent on the file name lists of the last 1000 commits, and
runtime was less than 0.5s for each of 998/1000 commits.  There's more a real
implementation might handle:

- pg_bsd_indent changes
- typedefs.list changes
- skip if the break-glass "pgindent: no" appears in a commit message
- commits changing so many files that a clean "git checkout" would be faster

> And it would have to also
> know to rebuild pg_bsd_indent on demand, which would require a full
> ./configure run (or meson equivalent). etc.
> 
> So while it might not be tens of seconds, it most definitely won't be fast.

A project more concerned about elapsed time than detecting all defects might
even choose to take no synchronous action for pg_bsd_indent and typedefs.list
changes.  When a commit changes either of those, the probability that the
committer already ran pgindent rises substantially.

Commits

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

  1. Fix comment from commit 22655aa231.

  2. Add a few recent commits to .git-blame-ignore-revs.

  3. Pre-beta2 mechanical code beautification.

  4. Pre-beta mechanical code beautification.

  5. Make agreed-on updates in perltidy options.

  6. Remove obsolete pgindent options --code-base and --build

  7. Integrate pg_bsd_indent into our build/test infrastructure.

  8. Sync pg_bsd_indent's copyright notices with Postgres practice.

  9. Import pg_bsd_indent sources.

  10. pgindent: filter files for the --commit option

  11. pgindent: more ways to find files to indent

  12. Fix pgindent --show-diff option.

  13. Add non-destructive modes to pgindent

  14. Initial pgindent run with pg_bsd_indent version 2.0.