svn commit: r476508 - head/x11/xorgproto

[Niclas Zeising]

On 08/11/18 11:23, Niclas Zeising wrote:
> On 08/06/18 14:20, Mathieu Arnold wrote:
>> Author: mat
>> Date: Mon Aug  6 12:20:39 2018
>> New Revision: 476508
>> URL: https://svnweb.freebsd.org/changeset/ports/476508
>>
>> Log:
>>    Fix Xprint build issue.
>>    For some reason, the Xprint.man patch is supposed to get into a 
>> subdirectory
>>    that does not exist:
>>      --------------------------
>>      |--- man/Xprint.man.orig    2018-07-24 10:46:56 UTC
>>      |+++ man/Xprint.man
>>      --------------------------
>>      (Creating file Xprint.man...)
>>      Patching file Xprint.man using Plan A...
>>      Empty context always matches.
>>      Hunk #1 succeeded at 1.
>>      done
>>    So patch extracts it in the first existing directory, and then the 
>> build fails:
>>      make[1]: don't know how to make man/Xprint.7. Stop
>>      make[1]: stopped in 
>> /wrkdirs/usr/ports/x11/xorgproto/work/xorgproto-2018.4
>>      *** Error code 1
>>      Stop.
>>    Sponsored by:    Absolight
>>
>> Modified:
>>    head/x11/xorgproto/Makefile   (contents, props changed)
>>
>> Modified: head/x11/xorgproto/Makefile
>> ============================================================================== 
>>
>> --- head/x11/xorgproto/Makefile    Mon Aug  6 11:23:41 2018    (r476507)
>> +++ head/x11/xorgproto/Makefile    Mon Aug  6 12:20:39 2018    (r476508)
>> @@ -14,4 +14,7 @@ CONFIGURE_ARGS= --without-fop \
>>           --without-xsltproc \
>>           --enable-legacy
>> +pre-patch:
>> +    @${MKDIR} ${WRKSRC}/man
>> +
>>   .include <bsd.port.mk>
>>
> 
> Strangely, I've never ran into this issue.  I created the patch pointing 
> to the directory, and it gets extracted, built and installed fine.  Are 
> there differences in how patch handles nonexistent directories on 
> dfifferent versions of FreeBSD?


Hi!
So, I just tried some more, in a bunch of different versions.  In all my 
jails, with poudriere, the man directory gets created and the patch is 
extracted there as it should be.  How are you building the port/package?
I'm just curious as to why we see this different behavior.  This went 
through an exp-run without issue, but something strange is happening here.
Regards
-- 
Niclas