On 2023-05-10 4:29 a.m., Stuart Henderson wrote:
On 2023/04/30 11:53, Antoine Jacoutot wrote:
+ if test "${enable_lua}" != "no"
+ then
+- PKG_CHECK_MODULES(LUA, lua5.2,
++ PKG_CHECK_MODULES(LUA, lua51,
Can we use SUBST_CMD to prevent hardcoding?
I don't see anything wrong with hardcoding here? I think MODLUA_VERSION
should reflect what the port wants, it's not like SHARED_LIBS.
On 2023/05/09 02:29, Brad Smith wrote:
On 2023-05-01 5:36 a.m., Stuart Henderson wrote:
Any particular reason to use 5.1 instead of 5.2? (They're all EoL except
5.4, but 5.1 is much longer since EoL).
Just that is what the default is, MODLUA_DEFAULT_VERSION.
I don't think this is meant to be prescriptive any more than
say AUTOCONF_VERSION, just that it's what most things used when the
module was added. We haven't run into problems with ports pulling
in multiple lua versions via lib deps.
(I have a diff to add lua5.X pkg-config files to the lua ports to
test sometime but the i386 builders are unhappy atm, also the include
directories are non-standard too and also need patches in various ports,
so it would be good to clean up that whole lot at once).
How about this diff for vlc? Compared to your previous:
- uses the lua version which upstream prefers
- brings back UPDATE_PLIST_ARGS from my earlier one to avoid the
a52/a${MODLUA_VERSION} that is subst'ed in PLIST-main with 5.2
I saw this part and did not understand what it was for.
- uses the minimum necessary configure.ac patch to handle lua52.pc.
Sure.
