* Tatsuki Makino <[email protected]> [20230817 17:14]: > Felix Palmen wrote on 2023/08/17 15:03: > > It is, when some file doesn't exist for all architectures, but for more > > than one of them. > > > > You first test the port on ARCH=foo. Then you test on ARCH=bar and find > > that it builds/installs an additional tmp/something, so you prefix it > > with %%BAR%% which expands to "@comment " unless building on ARCH=bar. > > > > Then you test on ARCH=baz which happens to build/install the same > > tmp/something. plist-check will still be happy because the file is there > > in pkg-plist (ignoring the @comment). Still, the file will be missing > > from the package. > > > > I might try another workaround, e.g. expand to "@comment dummy/" > > instead, so the name won't match and plist-check will complain ... > > epson-inkjet-printer-escpr2 is easy because it is restricted to 2 > architectures, but in general it seems difficult. > How about also making a combined PLIST_SUB in Makefile?
JFTR, this wasn't my problem here. In my concrete scenario, there are
only 3 supported archs anyways (I'm building a Linux-native toolchain
from source for the Linuxulator, will present it on this list once it
works on all 3 archs), and I already define NO_* subs, so it's straight
forward to set anything that might be required in pkg-plist.
The issue really was just about plist-check, which will *not* report
"orphaned" files when they are present in some @comment...
I now got around the issue using *this* block in my ports:
#v+
.include <bsd.port.options.mk>
.if ${ARCH} == amd64
PLIST_SUB+= AMD64="" NO_AMD64="@comment _nonexisting/"
.else
PLIST_SUB+= AMD64="@comment _nonexisting/" NO_AMD64=""
.endif
.if ${ARCH} == aarch64
PLIST_SUB+= AARCH64="" NO_AARCH64="@comment _nonexisting/"
.else
PLIST_SUB+= AARCH64="@comment _nonexisting/" NO_AARCH64=""
.endif
.if ${ARCH} == i386
PLIST_SUB+= I386="" NO_I386="@comment _nonexisting/"
.else
PLIST_SUB+= I386="@comment _nonexisting/" NO_I386=""
.endif
#v-
Sure, it's somewhat dirty workaround, but adding some "dummy" prefix to
the files makes sure plist-check will complain about them ;)
--
Felix Palmen <[email protected]> {private} [email protected]
-- ports committer -- {web} http://palmen-it.de
{pgp public key} http://palmen-it.de/pub.txt
{pgp fingerprint} 6936 13D5 5BBF 4837 B212 3ACC 54AD E006 9879 F231
signature.asc
Description: PGP signature
