| Age | Commit message (Collapse) | Author |
|---|
|
- make it more accurate by looking at files as they have been
committed (I often leave stuff out of commits)
- remove extraneous (I hope) parentheses
|
|
|
|
"(Un)tracked" rolls better off the tongue than "(un)committed". IMO.
I considered keeping track of "possibly lost pages" (i.e. uncommitted
files), but it's probably not an interesting metric since it will vary
from one computer to another.
The whole thing is not rigorous anyway, since tracked files might
contain uncommitted modifications.
Ah well. I still think it's a somewhat cute analogy.
|
|
Magit uses --literal-pathspecs when committing, so ls-files failed to
expand '*.md' in the word-counting script.
For a wild ride, I recommend going through Git's get_global_magic()
function in pathspec.c, which juggles with…
- GIT_FOO_PATHSPECS environment variables;
- --foo-pathspecs CLI arguments;
- ":(foo)" magic suffixes;
… where foo can be "literal", "glob", "noglob", or "icase".
|
|
(It's *critical* that I get the imaginary file hierarchy right before
I actually commit those notes, you see)
Make sure update-count.sh only looks at committed files; there's going
to be a lot of untracked files in there. Maybe I could add a
"possibly lost" count for those?
|
|
|
