Re: [PATCH] clean: new option --exclude-from

Jeff King [off-list ref] writes:

My motivation isn't exactly code sharing. It is that you sometimes want
to affect sub-commands of a program, and cannot pass command line
options to them yourself.

For instance, "git-stash --include-untracked" will call "git clean"
under the hood. There is no way to say "...and treat foo.* as ignored
for this invocation". It could grow its own "-e" option, but that does
not help any other third-party scripts which call "git clean".

So IMHO this is not really about command-line options, but about the
environment in which a command is executed. Environment variables are
the obvious way to do that; "git --foo" options are just syntactic sugar
to set the variables. We could just add variables without matching
options.

I'd be even more wary of that, as different commands use ignore
patterns for different purposes.  A script may invoke "clean" and
"add" and would want to use different sets of ignore patterns to
emulate "precious" class (which we do not have), for example, by
giving a wider ignore pattern for "add" to prevent a file that must
be kept untracked outside the index while telling "clean -X" that
that file is not expendable with a narrower ignore pattern.  That is
just one example that comes to me without thinking about the issues
too hard, so I am reasonably sure that it would hurt the ecosystem
to promote that the ignore pattern can be used for specifying
important per-invocation input to a script.

In any case, what we've been discussing may be an interesting and
potentially important tangent, but it still is a tangent while
evaluating the patch in question.  I do not think I'd be using the
new "--exclude-from=<file>" option to "clean" (simply because I do
not think I've ever used the existing "-e" option to it unless I am
merely testing the command to make sure it works as advertised)
myself, but I do not immediately see how it would hurt us in the
future to add it now.  So...