small errors in ghc 6.6

[Bulat Ziganshin]

Hello glasgow-haskell-users,

i've put my hands on the ghc 6.6/mingw32 dated 1st september
(http://www.haskell.org/ghc/dist/current/dist/ghc-6.5.20060901-i386-unknown-mingw32.tar.gz)

overall, compilation of my 5 kloc application was fine. on the
6.2->6.4 transition, i was bitten by two API changes, but this time
all APIs and language features i used remains stable. it's great!

the most funny part was that parallel arrays are still work :)))


nevertheless, i've captured several minor problems:

1. ghc +RTS -A10m aborts with the message:

ghc.exe: getMBlocks: VirtualAlloc MEM_COMMIT failed: The parameter is incorrect.

although i guess that it was already fixed since 1st September. btw,
how about adding using various RTS switches to ghc as test cases? GHC
is very large program and any problems with RTS flags would be captured in
this way


2. ghc compilation don't includes directory where Rts.h, Stg.h and
similar files resides, "$(GHC)\lib\i386-unknown-mingw32\include" on my
installation. it results in failing of via-C compilation, compilation
of *stub.c files, and inclusion of "Typeable.h". i was going around
this problem by using -I switch. it's possible that it's just
packaging problem. again - isn't it possible to catch this problem by
testsuite?

imho, that this error was not catched means that windows version isn't
really tested by anyone? how about trying to build various packages on
mingw32 version as test before releasing 6.6?


3. ghc 6.6 includes smart relinking capability which don't relink exe
file if it already exists and .hs source files was not changed. that's
great but ignores other .o files that can be also linked to program,
for example those that was compiled from C sources:

gcc -c -o a.o a.c
ghc --make Main.hs a.o

this command incorrectly don't relinks executable if a.c was changed
but Main.hs wasn't


-- 
Best regards,
 Bulat                          mailto:Bulat.Ziganshin at gmail.com