On Tue, Jun 17, 2008 at 4:36 PM, John Peacock
<john.peacock_at_havurah-software.org> wrote:
> Norbert Unterberg wrote:
>>
>> Who performs the CRLF to LF conversion during a commit, the server or
>> the client?
>
> The client code performs any conversions (from the local text-base, which is
> an exact representation of what the server sent). This is because there may
> be keyword substitution going on as well.
>
> The server normally stores the files as LF, but you can get into a state
> where files contain CRLF, but this is usually caused by external tools
> messing with the line endings when they shouldn't.
>
>> It seems that the 1.5 client sent the file including the CRLF to the
>> server instead of converting to LF during commit. A svn diff of the
>> changed file shows the eol-style change for each line. I still need to
>> verify that that CRLF are really present on the server.
>
> If the diff is showing that the EOL changed for each line, it is definitely
> the client that is at fault (but not necessarily the Subversion client). It
> could just have easily been Tortoise...
That is very strange. The svn diff was using the svn 1.4.3 command
line client. However, the commit was done using TSVN 1.5 rc3. I'll
better report the problem there, too.
Norbert
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-06-17 16:44:43 CEST