It does not correctly separate build/test/run depends or separate
dependencies per python version, but it does add dependencies it
thinks it needs and will recurse.
add it as is kmos@
diagnostics.
- show more info when @version is bumped, either updated info for simple
cases, or diffs when it's the only thing preventing an error.
- remove synthetic @version -1 from confusing diffs
- talk about missing version info when we end up with an error that we couldn't
solve.
okay sthen@
cores based on them.
Remove the localhost normal jobs/fetch jobs ambiguity by creating a dummy
"FETCH" host.
(still a bit of unwanted redundancy in hosts, but works better already)
version switch has changed dependencies for a fair sized chunk of packages on
all arches, and the gcc switch from 4.9->8 has changed dependencies for a huge
chunk of the package set on base-gcc arches.
* switch gcc4 module to gcc/8
* register the gcc/8 subpackages as updates for their gcc/4.9 counterparts
* sync the gcc version in devel/llvm
* add a quirk entry that gcj is obsolete
* unhook 4.9 from the build
Preparatory work by many.
ok sthen@
- if it wasn't there, just register it
- if it was there, don't allow it to go backwards
- when @version increases, anything goes. The new plist might have some
new objects, we don't really care.
as requested by sthen@
makes it possible to avoid REVISION bumps for large system changes (such
as gcc 4.9 -> 8.*)
it is advised to do a bulk with the newer register-plist first so that the
lists will have @version markers
- mention that COMMENT usually starts lower-case
- COMPILER for C++ ports
- mention use of CONFIGURE_STYLE=none to override default from a module
- example do-gen/BUILD_DEPENDS for autogen.sh-type ports
ok kn@, ok/typo fix pamela@
as is now done upstream by cpan tools. This is currently unused but in
perl 5.26+ adds "." to the module include path (@INC) as a large number
of cpan ports currently rely on this at build time.
espie@ agrees that this is sane.
Improves visibility of ports that are broken with LLD by using a common
variable rather than various workarounds, and allows simpler testing/debug
ld.bfd failures on amd64.
