Re: [PATCH 1/2] show-ref.c: Add missing call to git_config()

Junio C Hamano wrote:

Ramsay Jones [off-list ref] writes:

quoted

At present, 'git show-ref' ignores any attempt to set config
variables (e.g. core.checkstat) from the command line using
the -c option to git.

I think what you really want to see is not giving "-c" and have it
honored.

	"git show-ref" does not honor configuration variables at
	all, but at least core configuration variables read by
	git_default_config (most importantly core.checkstat) should
	be read and honored, because ...

would be more appropriate.  What are the variables that matter to
show-ref, and what are the reasons why they should be honored?  I
thought show-ref was just a way to enumerate refs, and does not read
the index nor checks if there are modifications in the working tree,
so I do not see any reason offhand for it to honor core.checkstat
(and I think that is the reason why we don't have the call there in
the current code).

:-D Yes, you caught me!

These patches *may* not be necessary, prior to Michael's
"reference related races" series. Specifically, the introduction
of the stat_validity_check() function to the reference handling API.

This means that the behaviour 'git show-ref' is now affected by
several config variables, including core.checkstat. I haven't
spent any time auditing the code, but the list would include
(at least) core.trustctime, core.filemode, core.checkstat,
core.ignorecygwinfstricks, ...

Exactly the same comment applies to 2/2.

ditto

Note that I am _not_ opposing these changes.  You brought them up
because you saw some reason why these should honor some core
variables.  I just want that reason to be explained in the log for
the future developers.

Yes, I will send a v2 (soon-ish, I hope).

ATB,
Ramsay Jones