GitLab

at least with -fno-pedantic-bottoms. This fixes #2915, and undoes some
of a522c3b2, on the grounds that with a flag `-fpedantic-bottoms`
around, we can be a bit more liberal when the flag is off..

Signed-off-by: Austin Seipp <austin@well-typed.com>

As usual, Mac OS X is extremely annoying (or the software is, anyway),
because not only does it load dynamic libraries with the .dylib
extension, but also the .so extension. For whatever reason. At least
it's easy to fix.
Signed-off-by: Austin Seipp <austin@well-typed.com>

As discussed in the ticket, after the landing of #8773, GND is now
-XSafe compatible.

This fixes the test fallout as well. In particular SafeLang07 was
removed following in the steps of SafeLang06, since it no longer failed
from GND, but failed due to roles and was thus invalid.

The other tests were tweaked to use TemplateHaskell instead of GND in
order to trigger safety warnings.
Signed-off-by: Austin Seipp <austin@well-typed.com>

For some reason on OS X, it seems like -Bsymbolic (which we use for
hooks into the RTS) isn't working, which results in #8754, where stats
don't work because defaultHooks doesn't initialize the stats flag. This
seems to work on Linux static/dynamically, but only on OS X statically.

After talking with Simon, really, the entire hooks thing is a bit
fragile. For now, we just work around it (since GHCi is dynamically
linked) by calling into the defaultHooks ourselves when GHC starts.
Signed-off-by: Austin Seipp <austin@well-typed.com>

This should have manifested earlier, but for some reason it only seemed
to trigger on Mavericks.
Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Disable -O2 optimization. Otherwise amount of generated C code makes
things very slow to compile (~5 minutes on core-i7 for 'compiler/hsSyn/HsExpr.lhs')

And sometimes not compile at all (powerpc64 overflows something
on 'compiler/hsSyn/HsExpr.lhs').

Issue #8748Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Austin Seipp <austin@well-typed.com>

Issue #8748Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Austin Seipp <austin@well-typed.com>

This fixes most of implicit function declarations emitted
C codegen in UNREG mode. Found by adding the following to
mk/build.mk:

    SRC_CC_OPTS += -Werror=implicit-function-declaration
    SRC_HC_OPTS += -optc-Werror=implicit-function-declaration

Issue #8748Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Austin Seipp <austin@well-typed.com>

UNREG mode has quite nasty invariant to maintain:
    capabilities[0] == &MainCapability

and it's a non-heap memory, while other
capabilities are dynamically allocated.

Issue #8748Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

To make a role annotation on a class asserting a role other than
nominal, you now need -XIncoherentInstances. See the ticket for
more information as to why this is a good idea.

  Which on OS X leaves macro unexpanded.

It validates and nofib shows no change, so possibly dead code. Removing in the
interest of code cleanliness, someone disagrees please revert (and preferably
add a testcase, or at least describe the situation this is important in in a
Note).

when matching a lambda in the template against an expression. When
matching, look through coercions (only for value lambdas for now), and
look through currently active unfoldings, if these are undersaturated,
i.e. produce a lambda.

This replaces the existing, somewhat fishy eta-expansion.

(This tests #2110.)

such as that of coerce.

just like boxed type equalities.

we want a rule "map coerce = coerce" to match the core generated for
"map Age" (this is #2110).

Some nice improvements on already succeeding test cases (#876, #7954
and #4267)

Test #149 needed a little change, lest call arity causes a allocation
change that we do not want to test here.

This also sets precedence for testing internals of GHC directly, i.e.
without trying to come up with Haskell code and observable effects.
Let's see how that goes.

I put all the tests (including those where the analysis could do better)
in one file because starting the GHC API is quite slow.

This analysis finds out if a let-bound expression with lower manifest
arity than type arity is always called with more arguments, as in that
case eta-expansion is allowed and often viable. The analysis is very
much tailored towards the code generated when foldl is implemented via
foldr; without this analysis doing so would be a very bad idea!

There are other ways to improve foldr/builder-fusion to cope with foldl,
if any of these are implemented then this step can probably be moved to
-O2 to save some compilation times. The current impact of adding this
phase is just below +2% (measured running GHC's "make").

Signed-off-by: Richard Eisenberg <eir@cis.upenn.edu>

Signed-off-by: Richard Eisenberg <eir@cis.upenn.edu>

We should still have pattern synonyms in TH, though.