From: Junio C Hamano <junkio@cox.net>
Xavier Maillard <zedek@gnu.org> writes:
> * Makefile: add git-blame as a candidate to the
> byte-compilation. Change the function to call to byte-compile.
Could we have a bit more detailed description on the latter
change?
Sure.
byte-compile-file is an interactive function. batch--byte-compile is a batch function.
First function would be (better) called when using emacs in its "interactive" form -ie. M-x byte-compile-file RET.
batch-byte-compile is reserved for batching use (no interaction with the user).
So in non interactive mode, this is prefered way to do things.
Explaining why something needed a change is the point of a
commit log message. Summarizing what got changed and how is
good but that is secondary, because anybody can see that you
changed the handcrafted --eval to -f by running "git show", even
if you did not have the above proposed log message.
Sorry Iam not "good" at commit log :/
What is left unsaid is _why_ it is better to use single dash
form of "-batch" and -f form instead of handcrafted --eval, and
the proposed log message does not clarify the reason.
First reason is given by C-h f batch-byte-compile RET which gives:
batch-byte-compile is an autoloaded Lisp function in `bytecomp'. (batch-byte-compile &optional noforce)
Run `byte-compile-file' on the files remaining on the command line. Use this from the command line, with `-batch'; it won't work in an interactive Emacs. Each file is processed even if an error occurred previously. For example, invoke "emacs -batch -f batch-byte-compile $emacs/ ~/*.el". If noforce is non-nil, don't recompile a file that seems to be
Second reason is "by habit" :)
Hope Ihave been clear. If you want me to submit something again from these patches, please tell how to do so (and modify what Ihave changed).
