Yup, -optl-u -optl Addr_Azh_con_info fixes it, and is better than putting HSlang.o in the compile as it will not automatically include the whole library in the binary. As for the slowdown, I will try and compare some code compiled with ghc-5.02 and I will let you know if anything real is happening here... Keean Schupke Department of Electrical & Electronic Engineering, Imperial College. -----Original Message----- From: glasgow-haskell-users-admin@haskell.org [mailto:glasgow-haskell-users-admin@haskell.org]On Behalf Of Simon Marlow Sent: 17 July 2002 11:38 To: k.schupke@ic.ac.uk Cc: glasgow-haskell-users@haskell.org Subject: RE: RTS interface followup...
rts_mkAddr is used by Linker.o in libHSrts.a (it looks like in the resolveObjs call - I in turn am calling this... (this was from doing an nm on the library)...
Hmm, yes the Linker does refer to it, just in case the dynamically linked program requires it. I think you should probably arrange that the missing symbols get linked in from HSlang.o, something like: ghc -package lang -optl-u -optl Addr_Azh_con_info does that help?
with regards to the threading, I was wondering if the default file buffering mode has changed, as text IO to the console seems slower (you can see characters 'printing' left to right, rather than just lines flying past)...
No, the default buffering has not changed. Can you quantify the slowdown, or provide an example? Does it only happen when --enable-threaded-rts is on? Cheers, Simon _______________________________________________ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users