ghc-5.00.2 -fvia-C: undeclared symbols

[Sigbjorn Finne]

hi,

you're running into trouble with some expeditious hacks that were
put into TclHaskell to make it work with ghc-4.xx. The following
change will make it compile with ghc-5.xx - have Check.h now
contain:

extern  StgTSO *run_queue_hd;
#if  __GLASGOW_HASKELL__ <= 408
extern StgTSO *MainTSO; /* temporary hack */
#define END_TSO_QUEUE  ((StgTSO *)(void*)&END_TSO_QUEUE_closure)
#endif

hth
--sigbjorn

----- Original Message -----
From: "Chris Dornan" <cdornan at arm.com>
To: <glasgow-haskell-bugs at haskell.org>
Sent: Monday, August 27, 2001 13:50
Subject: ghc-5.00.2 -fvia-C: undeclared symbols


> Hi,
>
> I am trying to compile TclHaskell (latest on haskell.org) with ghc-5.00.2
on
> Redhat 7.1 with an x86 thingie, but run into problems in trying to compile
a
> low-level module that calls out to C.
>
> ghc -fglasgow-exts -recomp -dcore-lint -syslib concurrent -fvia-C
'-#include
> "Check.h"' -c -ohi TclCompatibility.hi TclCompatibilityGhc.hs
> /tmp/ghc4365.hc: In function `s141_fast1':
> /tmp/ghc4365.hc:157: `END_TSO_QUEUE_closure' undeclared (first use in this
> function)
> /tmp/ghc4365.hc:157: (Each undeclared identifier is reported only once
> /tmp/ghc4365.hc:157: for each function it appears in.)
>
> I have noticed that ghc-5.00.2 seems to be more reliable with the x86
native
> generator but I don't think I can avoid it here as the module is calling
out
> to C.
>
> Is this a known problem?
>
> Can anyone suggest a work-around?
>
> Cheers,
>
> Chris Dornan
> cdornan at arm.com
>
>
>
> _______________________________________________
> Glasgow-haskell-bugs mailing list
> Glasgow-haskell-bugs at haskell.org
> http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs