[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
bug#31744: 26.1; Improvements to make tags and make -C test
From: |
Noam Postavsky |
Subject: |
bug#31744: 26.1; Improvements to make tags and make -C test |
Date: |
Sat, 09 Jun 2018 15:16:00 -0400 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux) |
Eli Zaretskii <eliz@gnu.org> writes:
>> Oh, hmm. The original reason, is that when looking at a source file in
>> Emacs, and then hitting M-. I get a prompt to visit the TAGS table,
>> which starts from the source directory. Then I have to go looking for
>> the TAGS file in the corresponding build directory.
>
> I have grown a habit a long time ago to "M-x visit-tags-table" before
> I issue the first "M-." command.
>
> The default prompt is fine, because what else would you expect Emacs
> to do in that case?
Yeah, there's no way for Emacs to know where the build directory is.
Maybe I'll add symlinks locally to work around this.
>> especially since every build directory will have identical TAGS
>> files anyway
>
> That's what happens currently, but it isn't carved in stone. We
> could, for example, have TAGS reflect only the files that are compiled
> in on the current platform; other projects (like GDB, for example) do
> just that. Then each build will have a different TAGS file.
I hope not, that sounds inconvenient to me.
> No, I think the principle is that the source tree holds everything
> that comes with a release tarball.
Okay, I dropped this change, and updated the FIXME comment to mention
this instead.
Going back to a question from earlier in the thread:
>> +ifeq ($(TEST_INTERACTIVE), yes)
>> + HOME=/nonexistent $(emacs) \
>> + -l ert ${ert_opts} \
>> + $(patsubst %,-l %,$(if $(findstring
>> $(TEST_LOAD_EL),yes),$ELFILES,$(ELFILES:.el=))) \
>> + $(TEST_RUN_ERT)
>> +else
>> -@${MAKE} -k ${LOGFILES}
>> - @$(emacs) -l ert -f ert-summarize-tests-batch-and-exit ${LOGFILES}
>> + @$(emacs) --batch -l ert -f ert-summarize-tests-batch-and-exit
>> ${LOGFILES}
>> +endif
> Not sure I understand the HOME trick: why not use -Q?
$(emacs) already includes -Q (or rather, the equivalent long options
--no-init-file --no-site-file --no-site-lisp). The HOME trick was added
in [d17aa3e535] and [412c38aa28]. I can't find a specific discussion,
but http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00641.html
seems related. I've added a comment to explain this better.
[412c38aa28]: 2017-05-30 08:39:39 -0700
Stop make check interacting with HOME
https://git.savannah.gnu.org/cgit/emacs.git/commit/?id=412c38aa28dd7e8363b481a09d1df62c40f9a5b7
[d17aa3e535]: 2017-05-30 12:50:54 -0400
Reduce scope of recent test/Makefile HOME change
https://git.savannah.gnu.org/cgit/emacs.git/commit/?id=d17aa3e535bba5e93ff188d5460c91001074255e
v2-0001-Make-tags-targets-respect-with-silent-rules-Bug-3.patch
Description: patch
v2-0002-test-Makefile.in-Add-TEST_INTERACTIVE-option-Bug-.patch
Description: patch
v2-0003-Reduce-quoting-for-SELECTOR-in-make-C-test-Bug-31.patch
Description: patch