First, temporary stop-gap measure: Just strip that symbol from
HSbase_cbits.o and hopefully you don't need it.
I've experienced something similar on Cygwin. There are four symbols
undefined that causes the exact same error (only with different symbols, one
is, for example, __imp___pctype).
WARNING! The rest is under-educated rampant speculation, take a shaker of
salt with this grain of thought.
I imagine the problem has to do with differing libraries linked in when
compiling as compared to interpreting. Either different libraries needed to
make an exe (my symbols seem related to importing definitions from mscrt in
MinGW) or "different" RTS's (your symbol looks like GMP stuff). As I don't
imagine GHCi is broken this blatantly in 5.05 (though try it and see) and it
would seem sensible for TH to use some of GHCi code, my suspicion is the
problem is in setup/interface code in TH for/to GHCi. Unless, of course,
GHCi is broken, then almost certainly that's the problem. Or perhaps
simply, the symbols just shouldn't be there.
Well, hopefully, you get a coherent response from someone in the know. In
the meantime, hopefully you'll make some work-around or my stop-gap solution
works for you. I'm not too gung-ho into looking into it myself; the
stop-gap solution works beautifully for me, and I can live with my program
exploding into glittery pieces if I use the wrong function (or more likely
simply not linking), since I doubt I'll ever use the ones related to the
"undefined" symbols I've stripped.
>From: Sean Seefried <sseefried(a)cse.unsw.EDU.AU>
>To: template-haskell(a)haskell.org
>Subject: [Template-haskell] Template Haskell bug
>Date: Sat, 15 Mar 2003 19:33:27 +1100 (EST)
>
>With GHC HEAD as of 14/03/2003 I get the following error when I try to
>compile:
>
>This was done on Debian Linux kernel version 2.4.20
>
>main = putStrLn (show ($([| 20 |])))
>
>---------------
>
>/home/sseefried/local/lib/ghc-5.05/HSbase_cbits.o: unknown symbol
>`__umoddi3'
>Loading package base ... linking ... ghc-5.05: panic! (the `impossible'
>happened, GHC version 5.05):
> can't load package `base'
>
>Please report it as a compiler bug to glasgow-haskell-bugs(a)haskell.org,
>or http://sourceforge.net/projects/ghc/.
>
>---------------
>
>Any idea why?
>
>Sean Seefried
>_______________________________________________
>template-haskell mailing list
>template-haskell(a)haskell.org
>http://www.haskell.org/mailman/listinfo/template-haskell
_________________________________________________________________
Add photos to your e-mail with MSN 8. Get 2 months FREE*.
http://join.msn.com/?page=features/featuredemail
With GHC HEAD as of 14/03/2003 I get the following error when I try to
compile:
This was done on Debian Linux kernel version 2.4.20
main = putStrLn (show ($([| 20 |])))
---------------
/home/sseefried/local/lib/ghc-5.05/HSbase_cbits.o: unknown symbol
`__umoddi3'
Loading package base ... linking ... ghc-5.05: panic! (the `impossible'
happened, GHC version 5.05):
can't load package `base'
Please report it as a compiler bug to glasgow-haskell-bugs(a)haskell.org,
or http://sourceforge.net/projects/ghc/.
---------------
Any idea why?
Sean Seefried
Urkle
I'm about to take off to Redmond for a week. Not sure when I'll get to
this,
but I hope I'll manage it sometime soon. I hope it's not a showstopper
for you.
(You might want to join the template-haskell mailing list)
simon
| -----Original Message-----
| From: Thomas Harke [mailto:harke@cse.ogi.edu]
| Sent: 10 March 2003 18:15
| To: glasgow-haskell-bugs(a)haskell.org
| Subject: ghc-5.05 / template haskell bug
|
| Hi,
|
| while experimenting with template haskell I encountered the following:
|
| ] ghc --make Main.hs
| ] Chasing modules from: Main.hs
| ] Skipping Power ( Power.hs, ./Power.o )
| ] Compiling Main ( Main.hs, ./Main.o )
| ] ghc-5.05: panic! (the `impossible' happened, GHC version 5.05):
| ] nameModule x {- v a1fV -}
| ]
| ] Please report it as a compiler bug to
glasgow-haskell-bugs(a)haskell.org,
| ] or http://sourceforge.net/projects/ghc/.
| ]
| ]
| ] make: *** [main] Error 1
|
| This was using a version very recently checked out of CVS (Friday, 14
March)
|
| The files that caused the problem are attached.
|
| --
| Tom Harke
| Computer Science and Engineering Department
| Oregon Graduate Institute
|
| Weiler's Law:
| Nothing is impossible for the man who doesn't have to do it himself
