instead of using runghc (the interpreter).
This is a *workaround* for the following two symptoms:
- Setup.hs: waitForProcess: resource exhausted (Resource temporarily
unavailable)
- Setup.hs: fd:15: hGetContents: illegal operation (Inappropriate
ioctl for device)
Note that this is not a *fix*. I'm sure there's some misbehaving
signal handler, either in runghc/ghci or in the ghc runtime library.
(No, it's *not* rts/posix/Signals.c:generic_handler(); I patched
it to preserve errno, but it didn't help)
Until I find the bug, it's better to let bulk builds pass without
(or with less) errors.
ok (with the workaround) espie@
into a common subdirectory `ghc', and also move the ghc distfiles
there (from ghc-${MODGHC_VER}).
This will avoid potential conflicts of hs-* ports distfiles coming
from hackage.haskell.org with distfiles used by other ports.
Based on an idea from dcoppa@ for databases/hs-redis.
ok dcoppa@
If you want to save some bandwdith and disk space, see my mail to
ports@, which contains a shell script that can move distfiles around
under your DISTDIR.
libraries not coming together with ghc. This allows for looking up
a library's PKGPATH by running
ghc-pkg field $pkgname pkgpath
where $pkgname is the GHC library name without the `hs-' prefix,
for example `ghc-paths'.
looks good to jasper@
cabal and without nort), to avoid collisions with other (non-Haskell)
ports;
- Documentation is installed as ${PREFIX}/share/doc/hs-$foo instead of
${PREFIX}/share/doc/$foo.
- The library itself (and its interface files) is installed as
${PREFIX}/lib/ghc/$foo instead of ${PREFIX}/lib/$foo.
- Additional files will be installed in ${PREFIX}/share/hs-$foo instead
of ${PREFIX}/share/foo.
from kili@
depending port's plists will be adjusted in a few
- Use integer-gmp again.
- Cleanout the extracted bootstrap directory right after installing it
to save some disk space.
- Use ${MAKE_ENV} instead ${MODGHC_SETUP_CONF_ENV} in ghc.port.mk (in
do-configure, use both).
- Don't compile Setup.l?hs, just use the interpreter (runghc) in
ghc.port.mk. This speeds up the build of most ports depending on
ghc and using a cabal-style build.
Necessary bumps and WANTLIB changes in ports using ghc will follow
later this evening.
just setting MODGHC_SETUP_CONF_ARGS (which is now empty by default).
Add dblatex-created documentation.
While here, use our INSTALL* macros where possible to get correct
permissions (noticed by dcoppa@). This does *not* fix the permissions
of libraries and interface files installed by Cabal-based tools,
because the permissions are hard-coded in Cabal, and I'm not going
to touch and fix Cabal ever, because IMHO it's completely broken
by design. (If you want to read some of this madness, have a look
at libraries/Cabal/Distribution/Compat/CopyFile.hs or even
libraries/Cabal/Distribution/Simple/Install.hs)
Expect some breakage of depending ports (at least of devel/darcs) and
some necessary WANTLIB changes, which will be fixed soonish.
Many thanks to Darrin Chandler and dcoppa@ for testing, reporting about
broken stuff, missing dependencies here and in ports depending on ghc.
Notes and rants:
- Bootstrapping is done using precompiled binaries, since .hc
bootstrapping still doesn't work. I really hate this.
THIS MEANS THAT GHC IS NOW AND WILL STAY LEGACY-ONLY (i386 and amd64)
At least until someone fixes it. I tried for more than two year
(well, only in my spare time and during my vacations) and failed.
- libgmp is currently disabled, because I didn't yet hack the GHC build
system to use the system libgmp instead of the patched one included
in GHC.
- The haddock ncluded in the ghc distfile is replaced by the version
of haddock found in devel/haddock. Haddock itself is @commented
in the ghc PLIST. Unfortunately, this needs an ugly hack that
introduces an otherwise useless pseudo flavor `no_deps' in
devel/haddock.
- CLDouble has been removed from GHC some time ago, because it was
an alias for double (AFAIK there's now support for long double
in GHC). As this isn't a really big problem, it currently breaks
c2hs, which I'll mark broken temporarily before committing the
ghc update.
- The external codeset defaults to latin1 (suggested by Simon Marlow)
and can be overridden by setting the HS_ENCODING to any codeset
supported by libiconv.
- ghc.port.mk still needs some love, especially for letting a port add
additional parameters to certain invocations of ${MODGHC_SETUP_PROG}.
