William Uther wrote:
>On 2/8/02 7:03 PM, "Blair Zajac" <blair@orcaware.com> wrote:
>
>
>
>>Greg Hudson wrote:
>>
>>
>>>On Thu, 2002-08-01 at 23:49, Blair Zajac wrote:
>>>
>>>
>>>>Why isn't this done for the native style? I don't see any need for it.
>>>>
>>>>
>>>If the repair flag is not set, the EOL translation is a reversible
>>>transformation; if you accidentally commit a binary file with
>>>svn:eol-style=native, then you can get the original contents back by
>>>checking it out again on the same platform.
>>>
>>>However, if the repair flag is set, the EOL translation is
>>>data-destroying.
>>>
>>>When we discussed this before (see the list archives--long thread,
>>>though), we were concerned about accidentally destroying data when
>>>svn:eol-style=native, but felt that fixed eol-styles were specialized
>>>enough that they were very unlikely to accidentally appear on binary
>>>files.
>>>
>>>
>>Well, it's in the tree now in rev 2864.
>>
>>I think we should look at the problems people are actually seeing. There
>>was one problem with a checkin this week on RapidSVN and problems I've been
>>having at work (we are now using svn!!!) where people are doing commits of
>>mixed eol text files.
>>
>>
>
>I like the non-data-destroying guarantee. Maybe have --force fix the line
>endings? And a note in the error to use --force to override?
>
This seems to be the accepted way of doing things, yes. +1 on skipping
the tests with --force. -0.5 on always skipping the tests.
Blair, that -0.5 may change to -1 the moment I bork a binary file of
mine. ;-)
--
Brane Čibej <brane_at_xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Aug 5 02:42:06 2002