Re: What about Perl autodie?

From: Peter Eisentraut <peter(at)eisentraut(dot)org>
To: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Daniel Gustafsson <daniel(at)yesql(dot)se>, John Naylor <johncnaylorls(at)gmail(dot)com>, Greg Sabino Mullane <htamfids(at)gmail(dot)com>
Subject: Re: What about Perl autodie?
Date: 2024-03-18 06:27:56
Message-ID: 75cab15d-1c67-4110-b094-1aa84aa2c470@eisentraut.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 21.02.24 08:26, Peter Eisentraut wrote:
> On 14.02.24 17:52, Peter Eisentraut wrote:
>> A gentler way might be to start using some perlcritic policies like
>> InputOutput::RequireCheckedOpen or the more general
>> InputOutput::RequireCheckedSyscalls and add explicit error checking at
>> the sites it points out.
>
> Here is a start for that.  I added the required stanza to perlcriticrc
> and started with an explicit list of functions to check:
>
> functions = chmod flock open read rename seek symlink system
>
> and fixed all the issues it pointed out.
>
> I picked those functions because most existing code already checked
> those, so the omissions are probably unintended, or in some cases also
> because I thought it would be important for test correctness (e.g., some
> tests using chmod).
>
> I didn't design any beautiful error messages, mostly just used "or die
> $!", which mostly matches existing code, and also this is
> developer-level code, so having the system error plus source code
> reference should be ok.
>
> In the second patch, I changed the perlcriticrc stanza to use an
> exclusion list instead of an explicit inclusion list.  That way, you can
> see what we are currently *not* checking.  I'm undecided which way
> around is better, and exactly what functions we should be checking.  (Of
> course, in principle, all of them, but since this is test and build
> support code, not production code, there are probably some reasonable
> compromises to be made.)

After some pondering, I figured the exclude list is better. So here is
a squashed patch, also with a complete commit message.

Btw., do we check perlcritic in an automated way, like on the buildfarm?

Attachment Content-Type Size
v2-0001-Activate-perlcritic-InputOutput-RequireCheckedSys.patch text/plain 15.7 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrew Dunstan 2024-03-18 06:34:23 Re: What about Perl autodie?
Previous Message Michael Paquier 2024-03-18 06:13:41 Re: Weird test mixup