Re: [PATCH 1/1] cvsexportcommit: force crlf translation

From: Junio C Hamano <gitster@pobox.com>
Date: May 8, 2019, 10:59 UTC
Message-ID: <xmqqr299fbyw.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <nycvar.QRO.7.76.6.1905081248190.44@tvgsbejvaqbjf.bet>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

Show 13 quoted lines

> How about this:
>
> 	The offical CVS for Windows (called CVS NT) produces DOS line
> 	endings in its `cvs status` output. Let's teach our own
> 	`cvsexportcommit` command to handle that gracefully.
>
> It is unlikely that anybody wants to spend time "fixing" this in CVS NT,
> even less likely that anybody would take that patch, and even if that was
> the case, there will still be plenty of CVS NT versions out there that
> `cvsexportcommit` cannot handle.
>
> I think it would be best to just integrate this change in Git and be done
> with it. It's not like it adds a ton of maintenance burden there.

One thing that took some time (at least to me) to audit was that the helper being touched is *NOT* limited to running "cvs status". If other commands cared about passing binary data intact, this change would have been disasterous. After seeing what external commands it is told to drive, I think it is OK. It's not like we'd be updating the program often and reusing the function blindly to accept possibly binary data in a near future.