If we use a --docdir that's not specific to the package, then different builds
will install their license files into the same location, which leads to file
collisions if those are ever joined into the same environment.
Fixes https://github.com/NixOS/nixpkgs/issues/35024.
enableShared in generic-builder.nix should default to what the GHC
compiler was compiled with. Add a passthru to all of the GHC compilers
to hold the value of enableShared. If enableShared is not set in the
GHC we just use false as the default value for enableSharedLibraries.
Note: I may have missed some compilers. Only GHC & GHCJS are covered
by this commit but this shouldn’t break evaluation of anything else.
This decreases complexity and ensures setup dependencies are properly
specified with `setup-depends` as they should be. Testing will say if
this is a reasonable change.
The reason why this does not work is not that we can't built static
objects, we can, but we can't use `-staticlib` on GHC on windows.
`-staticlib` rolls all dependencies into a combined archive. While this
would work on windows if we used gnu ar and MRI script, GHC can't rely
on GNU ar, and as such has a quick archive concatenation module for GNU
and BSD archives only.
This is because they are just for Setup.hs, so they are just used at build time
and completely isolated from the normal components' dependencies.
This was previous implemented in 8a8f0408cd, but
reverted in e69c7f5641 because it broken
setup-depends non-cross in haskell shell environments (custom Setup.hs in cross
shell environments has never worked). This version adds a special native
exception to avoid that breakage.
Just like with the other `--extra-*` flags, cc/ld-wrapper already handles
this, but we need to make Cabal aware so that the haskell builds have
the correct metadata.
The reason why this does not work is not that we can't built static
objects, we can, but we can't use `-staticlib` on GHC on windows.
`-staticlib` rolls all dependencies into a combined archive. While this
would work on windows if we used gnu ar and MRI script, GHC can't rely
on GNU ar, and as such has a quick archive concatenation module for GNU
and BSD archives only.
This is because they are just for Setup.hs, so they are just used at build time
and completely isolated from the normal components' dependencies.
This was previous implemented in 8a8f0408cd, but
reverted in e69c7f5641 because it broken
setup-depends non-cross in haskell shell environments (custom Setup.hs in cross
shell environments has never worked). This version adds a special native
exception to avoid that breakage.
Just like with the other `--extra-*` flags, cc/ld-wrapper already handles
this, but we need to make Cabal aware so that the haskell builds have
the correct metadata.
Following legacy packing conventions, `isArm` was defined just for
32-bit ARM instruction set. This is confusing to non packagers though,
because Aarch64 is an ARM instruction set.
The official ARM overview for ARMv8[1] is surprisingly not confusing,
given the overall state of affairs for ARM naming conventions, and
offers us a solution. It divides the nomenclature into three levels:
```
ISA: ARMv8 {-A, -R, -M}
/ \
Mode: Aarch32 Aarch64
| / \
Encoding: A64 A32 T32
```
At the top is the overall v8 instruction set archicture. Second are the
two modes, defined by bitwidth but differing in other semantics too, and
buttom are the encodings, (hopefully?) isomorphic if they encode the
same mode.
The 32 bit encodings are mostly backwards compatible with previous
non-Thumb and Thumb encodings, and if so we can pun the mode names to
instead mean "sets of compatable or isomorphic encodings", and then
voilà we have nice names for 32-bit and 64-bit arm instruction sets
which do not use the word ARM so as to not confused either laymen or
experienced ARM packages.
[1]: https://developer.arm.com/products/architecture/a-profile
Since GHC is a cross compiler, it's perfectly possible to make haskell
binaries on platforms without GHCs. `windows ++ unix` seems good enough
for now.
Also don't default `hydraPlatforms` to `platforms`. The former must be a
list of systems (strings), but the latter is a list of systems or
patterns.
This change raises the question of whether we should disable building of shared
libraries altogether (since we don't link them). Unfortunately, we can't,
because GHC expects shared libraries to be around when building and running
test suites, and no amount of passing --disable-shared flags at configure time
changes the fact. I *guess* that's related to the DYNAMIC_GHC_PROGRAMS = YES
flag we set when building GHC itself, but I haven't investigated it further
yet.
Closes https://github.com/NixOS/nixpkgs/issues/29011.
Unless specified otherwise, the default package set compiles library variants
for profiling with profiling-detail "all-functions". Executables, however, are
not built with profiling enabled.
This change increases the closure size for many Haskell programs, but the
practical advantage of having stack traces and performance measurements easily
available during development seems to outweigh that cost.
Closes https://github.com/NixOS/nixpkgs/issues/22340.
This reverts commit e73e5c884f. Please don't
set $LD_LIBRARY_PATH! Instead, pass appropriate --extra-include-dir and
--extra-lib-dir arguments to "cabal configure" to ensure that Cabal knows
about system dependencies.
For a Haskell package "foo" the environment foo.env now contains the build tool
dependencies required for compiling the package in $PATH.
Fixes https://github.com/NixOS/cabal2nix/issues/331.
Cabal 1.x says:
| Warning: This package indirectly depends on multiple versions of the same
| package. This is highly likely to cause a compile failure.
But in version 2.x, that warning is split into two lines differently:
| Warning:
| This package indirectly depends on multiple versions of the same package. This is very likely to cause a compile failure.
This commit modifies the call to "egrep" to recognize both versions by virtue
of the "-z" flag, which essentially interprets the whole configure-time output
as one long line.
This works by extracting out some logic from generic-builder.nix to
make it possible to get at the relevant information by overriding
mkDerivation for the haskell package.
This partially undoes the change from 8788bfe762.
The 'doBenchmark' name is more consistent with the naming scheme used for
other phases, like 'doCheck', 'doHaddock', etc.
This reverts commit dfb0f25484, reversing
changes made to 7f8ff02437. These changes broke
the ghcWithPackages wrapper:
nix-shell -p "haskellPackages.ghcWithPackages (ps: [ps.mtl])" --run "ghc-pkg list mtl"
/nix/store/szz84j5k1dy3jdashis6ws28d8l8zxxb-ghc-8.0.2-with-packages/lib/ghc-8.0.2/package.conf.d
(no packages)
WIP
If the `hoogle.nix` file wants to have any sane chance of finding haddock
outputs, the packages need to export the haddock folders as an identifier.
A few TODOs still stand, like passing self to the package instead of needing to
pass it to `haddockDir`. Maybe the exported identifier should be integrated into
the fixpoint somehow instead of using `passthru`?
We achieve this by moving setupHaskellDepends from the buildInputs attribute
into "otherBuildInputs", which is the attribute the builder uses to construct
the build inputs in both the actual build as well as the "env" attribute.
As @oxij points out in [1], this breakage is especially serious because
it changes the contents of built environments without a corresonding
change in their hashes. Also, the revert is easier than I thought.
This reverts commit 3cb745d5a6.
[1]: https://github.com/NixOS/nixpkgs/pull/27427#issuecomment-317293040
This makes those files a bit easier to read. Also, for what it's worth,
it brings us one baby step closer to handling spaces in store paths.
Also, I optimized handling of many transitive deps with read. Probably,
not very beneficial, but nice to enforce the pkg-per-line structure.
Doing so let me find much dubious code and fix it.
Two misc notes:
- `propagated-user-env-packages` also needed to be adjusted as
sometimes it is copied to/from the propagated input files.
- `local fd` should ensure that file descriptors aren't clobbered
during recursion.
This actually will matter when I (soon) land cross-compilation support,
and native deps are compiled for the build, instead of host, platform.
But even now, it's good style do this, and one less thing to review
later.
The use of `$nativePkgs` is unfortunate, as it will need to be
swapped out for `$crossPkgs` for build != host builds. This will just a
temporarily cause pain, however, as eventually I will change stdenv to
use a `$runPkgs` (or moral equivalent) in both case.
Closes https://github.com/NixOS/nixpkgs/pull/26480.
By default, `ghcjs` haddocks are precluded from being built. I see no reason to disallow building haddocks on projects built with `ghc` and `ghcjs` (someone can correct me here). `HaLVM` currently does not support `haddock` since it's a Stage 1 `GHC`. https://github.com/GaloisInc/HaLVM/blob/master/src/misc/build.mk.in#L20
Currently, building `haddocks` for `ghcjs` projects requires altering the derivation, which doesn't allow users to take advantage of the cache. This change will relieve that.
As noted in #25595, a change introduced in 4b77d425aa causes an
explosion of inodes due to the constructions of many, many `ghcEnv`
symlink forests. This commit undoes that change.
To discuss reworking the support for GHCJS plugins, please see: #26192Fixes#25595
The Sierra linker added a limit on the number of paths that any one
dynamic library (`*.dylib`) can reference. This causes problems when
a Haskell library has many immediate dependencies (#22810).
We follow a similar fix as GHC/Cabal/Stack: for each derivation,
create a new directory with symlinks to all the dylibs of its immediate
dependencies, and patch its package DB to reference that directory
using the new `dynamic-library-dirs` field.
Note that this change is a no-op for older versions of GHC, i.e., they will
continue to fail on some packages as before.
Also note that this change causes the bootstrapped versions of GHC to be
recompiled, since they depend on `hscolour` which is built by
`generic-builder.nix`.
Tested by building the `stack` binary as described in #22810.
