Thread: Modest proposal: run check_keywords.pl on every build

Modest proposal: run check_keywords.pl on every build

From
Tom Lane
Date:
I've had it with mopping up after oversights like this one:
http://archives.postgresql.org/pgsql-hackers/2012-09/msg01057.php

We made a script (src/tools/check_keywords.pl) to check for this type of
error years ago, and even added it to the top-level "maintainer-check"
target awhile back, but nonetheless people continue to screw it up on
a regular basis (twice already this year alone).  I think we should move
the script into src/backend/parser and run it as part of the make gram.c
target.  (Doing it that way will not cause perl to become required for
building from tarballs, since gram.c is supplied prebuilt in tarballs.
And we already require perl for builds from raw git pulls.)

Any objections?
        regards, tom lane



Re: Modest proposal: run check_keywords.pl on every build

From
Andrew Dunstan
Date:
On 09/26/2012 09:16 PM, Tom Lane wrote:
> I've had it with mopping up after oversights like this one:
> http://archives.postgresql.org/pgsql-hackers/2012-09/msg01057.php
>
> We made a script (src/tools/check_keywords.pl) to check for this type of
> error years ago, and even added it to the top-level "maintainer-check"
> target awhile back, but nonetheless people continue to screw it up on
> a regular basis (twice already this year alone).  I think we should move
> the script into src/backend/parser and run it as part of the make gram.c
> target.  (Doing it that way will not cause perl to become required for
> building from tarballs, since gram.c is supplied prebuilt in tarballs.
> And we already require perl for builds from raw git pulls.)
>
> Any objections?
>
>             


works for me.

cheers

andrew



Re: Modest proposal: run check_keywords.pl on every build

From
Robert Haas
Date:
On Wed, Sep 26, 2012 at 9:16 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Any objections?

None, and my apologies for the oversight.

I would really like to have a cross-check for duplicated OIDs, too.
After a couple of years I think I have learned most of the rules for
what stuff I need to cross-check before committing, but it is a
nuisance to remember to do it each time and, despite my best efforts,
I have fouled things up a few times.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company