| From: | Peter Eisentraut <peter(at)eisentraut(dot)org> | 
|---|---|
| To: | Tristan Partin <tristan(at)neon(dot)tech> | 
| Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> | 
| Subject: | Re: make dist using git archive | 
| Date: | 2024-02-13 06:53:57 | 
| Message-ID: | 34bfa53d-671d-4592-8d7a-26e51c3ed45e@eisentraut.org | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-hackers | 
On 12.02.24 18:26, Tristan Partin wrote:
> On Sun Feb 11, 2024 at 5:09 PM CST, Peter Eisentraut wrote:
>> Small update: I noticed that on Windows (at least the one that is 
>> running the CI job), I need to use git -c core.autocrlf=false, 
>> otherwise git archive does line-ending conversion for the files it 
>> puts into the archive.  With this fix, all the archives produced by 
>> all the CI jobs across the different platforms match, except the 
>> .tar.gz archive from the Linux job, which I suspect suffers from an 
>> old git version.  We should get the Linux images updated to a newer 
>> Debian version soon anyway, so I think that issue will go away.
> 
> I think with this change, it is unlikely I will be able to upstream 
> anything to Meson that would benefit Postgres here since setting this 
> option seems project dependent.
Meson is vulnerable to the same problem: If the person who makes the 
release had some crlf-related git setting activated in their 
environment, then that would affect the tarball.  And such a tarball 
would be genuinely broken for non-Windows users, because at least some 
parts of Unix systems can't process such CRLF files correctly.
(This is easy to test: Run meson dist with core.autocrlf=true on the 
postgresql tree on a non-Windows system.  It will fail during dist check.)
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Joel Jacobson | 2024-02-13 07:05:03 | Re: Possibility to disable `ALTER SYSTEM` | 
| Previous Message | Amit Kapila | 2024-02-13 06:50:30 | Re: Synchronizing slots from primary to standby |