I'm moving a discussion from the bug tracker (http://bugs.darcs.net/ 
issue538) to darcs-users:

Op 25-apr-2008, om 13:05 heeft Eric Kow het volgende geschreven:
>
>>  Hell, the existing --set-scripts-executable switch as used by  
>> apply is
>>  already a hack -- you might as well rip that out and replace it with
>>  the aforementioned utility.
>
> Right, that's what I meant.

--set-scripts-executable is a hack indeed. I made a simple fix for  
the darcs-1 repository case, but - again - how to scale it to hashed  
repositories wasn't obvious. In a darcs-1 repository, I can store the  
test script in the pristine tree with the executable bit set and all  
is well. But in a hashed repo, the pristine tree isn't stored as a  
directory in the filesystem, and the format of the pristine in hashed  
repos seems not to specify whether a file is executable or not. So I  
would either have to extend the hashed pristine format to include  
executability, or to change the semantics of --set-scripts-executable  
on darcs-1 repos, or just accept different behavior of --set-scripts- 
executable on different repo formats.

I suspect that this difference in semantics also shows with darcs  
get: if you darcs get a darcs-1 repo with --set-scripts-executable,  
your test script should work; if you get a hashed repo, it won't.

If you know more about the intended semantics of darcs than I do:  
what option should I choose here? Are there any compelling reasons  
for keeping --set-scripts-executable as a darcs flag instead of an  
external tool? BTW, how do other VCSs handle executability (it seems  
that svn stores executability in the pristine)?

Regards,
Reinier

PS: twb: Maybe an even simpler workaround would be to use 'sh  
mytest.sh' as the test command instead of './mytest.sh'.


  
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to