There is no file_isdir() handling in the ci/commit code, and even more confusingly, the object-types and style seem different between the two (add and ci/commit) blocks.

Could these share a common function that recurses into dirs and returns strings/objects that are the files? Could fossil ci * be made to behave same as fossil add *?

drh added on 2009-02-16 12:55:35:
Simply omit the "*" and "fossil commit" automatically commits everything. Why do you want to include "*" syntax?

bharder added on 2009-02-16 14:45:03:
If I have a logical group of commits in a single dir, and don't want to commit every single outstanding commit, then "*" is useful.

kkinnell added on 2009-02-20 22:54:17:
On a Unix or GNU box you can do this if

• You're not trying to recurse

• There aren't any 'extra' files in the directory

• You get the full path from the root of the checkout.

That last is as easy as starting at the checkout root, and
    fossil commit path/to/the/checkins/*

You may not even need the "full" path—but it never hurts to over-specify a path.

A short Perl (or some lesser language ;) script can clean and commit a particular directory in one swell foop.

anonymous added on 2009-03-14 23:45:46:
You can fix all kinds of problems by wrapping them in Perl, ultimately down to the point of writing an entire SCM in it. I don't think asking for consistency in behavior is that unrealistic a request.

stephan added on 2009-11-30 14:44:18:
Keep in mind that the '*' is parsed by the SHELL, and NOT by fossil. What fossil gets as input is whatever the shell expanded '*' to be.