GitLab

This is a fix for Trac #7101

First, make Rules.match_co able to deal wit some modest coercions
Second, make SpecConstr use wild-card for coercion arguments

This is the rest of the fix for Trac #7165

This is part of the fix to Trac #7165

The constraint solver doesn't zonk the types of coercion variables
so we can't assume that a coercion variable will have a (~) type.

Fixes Trac #7090.

Reported by Ganesh, Trac #7178.  Fix is easy.

Simon Marlow spotted that we were #include'ing MachRegs.h several times,
but that doesn't work as (a) it uses ifdeffery to avoid being included
multiple times, and (b) even if we work around that, then the #define's
from previous inclusions are still defined when we #include it again.

So we now put the platform code for each platform in a separate .hs file.

Peter Wortmann authored Aug 08, 2012 and Simon Marlow committed Aug 21, 2012

I suppose this was a good idea for HPC, as it assumed that source code
annotations coming from a source file could only talk about the same
source file (by how Mix files are saved).

I don't see a reason why cost-centres or source annotations would want
that kind of behaviour. I introduced a flag for toggling the behaviour
per tickish.

(plus some minor refactoring, as well as making sure that the same check
applies to binary tick boxes, where they had apparently been forgotten.)

See Note [Impedence matching] in TcBinds.
Fixes Trac #7173

Fixes Trac #7164

Erik de Castro Lopo authored Aug 08, 2012 and Simon Marlow committed Aug 21, 2012

Code that needs to differentiate between i386 and x86-64 should now
be written as if x86-64 is the default and i386 is the special case.
Eg:

    # if i386_TARGET_ARCH
    someFuncion = .....
    # else
    someFuncion = .....
    # endif

The calculation should be done in one place, of course.