[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Windows buildbot failures

From: Branko Cibej <brane_at_xbc.nu>
Date: Fri, 03 Apr 2009 16:11:44 +0200

Arfrever Frehtes Taifersar Arahesis wrote:
> 2009-04-03 13:43 Bert Huijben <rhuijben_at_sharpsvn.net> napisaƂ(a):
>
>>> -----Original Message-----
>>> From: Bert Huijben [mailto:rhuijben_at_sharpsvn.net]
>>> Sent: donderdag 2 april 2009 22:47
>>> To: dev_at_subversion.tigris.org
>>> Subject: Windows buildbot failures
>>>
>>> Hi,
>>>
>>> Thanks to Paul the reporting of the Windows tests is fixed, so we are now
>>> back to the following list of failed tests (on the Windows buildbot and
>>> confirmed here locally):
>>>
>>> At least one test FAILED, checking M:\svn-auto-test\fsfs\tests.log
>>> FAIL: revert_tests.py 1: revert relative to wc root
>>> FAIL: patch_tests.py 1: 'svn patch' basic functionality with no unidiff
>>> FAIL: patch_tests.py 2: apply a unidiff patch -- test external tool
>>> FAIL: patch_tests.py 3: test copy and move operations
>>>
>>> I think those last 3 are still related to the merging of the svn-patch
>>> branch?
>>> The XFail of these tests was removed in r36564 (two weeks ago). I don't
>>> think the patch detection works corrently on these tests.
>>>
>> These 3 tests are skipped if no patch is found, but the Windows buildbot and
>> the PC I used for testing both have a patch.exe.
>>
>> The issue I see now is that the created patches (in the patch test) have
>> "\r\n" (Windows style) newlines and it looks like our svn patch expects "\n"
>> (Unix style) newlines.
>>
>> Changing the newlines of the generated patch to "\n" resolved the "patch:
>> **** Only garbage was found in the patch input." error when used outside the
>> test framework, but I think there are more problems than just this one.
>>
>> I think we should at least allow "\r\n" newlines in the non-unified-patch
>> blocks (The 'SVNPATCH1 BLOCK' blocks) on systems that don't default to "\n".
>>
>
> Sometimes Windows users send patches to Unix users, so both "\r\n" and
> "\n" should be allowed on all systems. Feel free to implement it :) .
>

It is actually quite wrong to expect any kind of specific line endings
in a patch, because when the patch is applied, any changed line endings
*do* become part of the patched file.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=1532635
Received on 2009-04-03 16:12:04 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.