[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [lmi] PATCH: Assorted fixes for the CI builds
From: |
Greg Chicares |
Subject: |
Re: [lmi] PATCH: Assorted fixes for the CI builds |
Date: |
Mon, 13 Jun 2022 13:56:52 +0000 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Thunderbird/91.8.0 |
On 6/12/22 22:14, Vadim Zeitlin wrote:
>
> I'd like to merge https://github.com/let-me-illustrate/lmi/pull/217
Ready to push after testing.
> I hesitated to include a similar change to nychthemeral_test.sh, where it
> is also clearly going to be needed once you update to Wine 7, but I'm not
> set up to actually test it, so I didn't do it -- please let me know if
> you'd like me to do it or if you'll do it yourself.
Debian 'bookworm' still has wine-6.0.3~repack-1, so I can't test it
(unless I were to start using 'sid' as an alternative, but I don't
really want to do that now). But I'll make speculative changes, and
then test their sufficiency when I upgrade to wine-7, unless you want
to test them first.
Right now, only makefiles contain a definition of $(EXEEXT), but wine-7
apparently requires scripts to have a definition, too. I believe that it
should be handled the same way $PERFORM is handled, i.e., it should be
set OAOO in 'set_toolchain.sh' and then referenced as environment
variable $EXEEXT in scripts wherever $PERFORM is used.
IOW, we seek to establish that this command:
git grep '\<PERFORM\>[^X]*$'
will find only 'export' or assignment statements. (The [^X] thing is
a simplistic negative lookahead for "EXEEXT".)
> I see that it's also
> needed for
>
> - XMLLINT definition in msw_generic.make
Yes, but, I believe, only for the commented-out definition:
# Instead of requiring installation of the build system's own libxml2:
XMLLINT := xmllint
# use the one that lmi builds:
-# XMLLINT := $(PERFORM) $(localbindir)/xmllint
+# XMLLINT := $(PERFORM) $(localbindir)/xmllint$(EXEEXT)
# but don't do that until 'wine' is fixed--see:
# https://lists.nongnu.org/archive/html/lmi/2019-01/msg00034.html
> - In test_coding_rules_test.sh (but not the concinnity target in the main
> makefile, as TEST_CODING_RULES already includes the extension)
Actually, because that binary cannot be built for MinGW-w32:
@[ "$$LMI_TRIPLET" = "x86_64-pc-linux-gnu" ] \
|| ($(ECHO) "'$@' requires x86_64-pc-linux-gnu." && false)
@$(MAKE) test_coding_rules$(EXEEXT)
@$(INSTALL) -c -m 0775 $(TEST_CODING_RULES) $(localbindir)
there's also no need for that script to work with 'wine'. In this case,
the right way to follow the "$PERFORM requires $EXEEXT" rule above is
to remove "$PERFORM" from that script.
> - For generate_passkey in workhorse.make
Indeed.
> Again, please let me know if you'd like me to add the corresponding
> changes to this branch or if you'd rather make them yourself.
Done; tested; pushing after I hit "Send".