Currently, Darcs will implicitly add the directory containing a to-be-added file, if it is not known by Darcs. This is desirable, but has shown up some strange bugs.
Particularly, adding a non-existent file within a newly-added directory will cause Darcs to print an error message, but it will exit successfully. Something like: $ mkdir subdir $ cd subdir $ darcs add foo File subdir/foo does not exist! $ echo $? 0 $ darcs wh adddir ./subdir $ darcs add foo File subdir/foo does not exist! darcs failed: No files were added $ echo $? 2 The second time add is called, Darcs exits with a non-zero status, since the directory has already been added. The current fix I have coded will only add files/directories if all the files specified exist (if any fail to be added, no changes will be made). However, this could be somewhat heavy-handed - someone might want to specify a long list of possibly-existing files, and tolerate the failures. I can't think of an instance of this use-case (mainly because I tend to add files using shell-globbing, which wouldn't cause a non-existent name to be specified), but I'd like to know if anyone has used something that is similar, where my change would break the expected behaviour. What should Darcs do in the case where non-existent files are added? Cheers, Owen.
_______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
