On 30 October 2015 at 13:40, Bert Huijben <bert_at_qqmail.nl> wrote:
>> -----Original Message-----
>> From: Ivan Zhakov [mailto:ivan_at_visualsvn.com]
>> Sent: vrijdag 30 oktober 2015 10:52
>> To: Branko Čibej <brane_at_apache.org>
>> Cc: dev_at_subversion.apache.org
>> Subject: Re: svn commit: r1703689 - in /subversion/trunk/subversion:
>> libsvn_client/merge.c tests/cmdline/merge_automatic_tests.py
>
>
>> >>> + SVN_ERR(svn_stream_open_readonly(&mine_stream,
>> mine_abspath,
>> >>> + scratch_pool, scratch_pool));
>> >>> +
>> >>> + if (!special && (eol_style || keywords))
>> >> Hi Bert,
>> >>
>> >> Do you know why we do not convert eol style/collapse keywords for
>> >> special files? I know that it's not related to this commit, but may be
>> >> you know rationale behind this behavior.
>> >
>> > The contents of special files are not user-defined and semantically, the
>> > contents aren't text but control strings interpreted by the Subversion
>> > client. There's a pretty good case for not allowing svn:eol-style or
>> > svn:keywords properties on such files in the first place.
>> >
>> This is makes sense. Thanks for the explanation.
>
> I have nothing to add to the previous comments. Thanks Branko and Ivan.
>
>
> Maybe I should note that I reviewed other similar code when I applied this patch in Berlin. There are a few not 100% correct cases around 'svn cat' and blame, but non that can really affect our behavior.
> (And all of these worked the same way since at least 1.5.x)
>
I suspected the same. I wondered that this logic does not exists as
some API function. May be it worth to factor out this code? I'm going
to put my +1 for backport anyway.
--
Ivan Zhakov
Received on 2015-10-30 12:10:57 CET