***ChanServ sets mode: +o oriansj
***ChanServ sets mode: +o janneke_
***attila_lendvai_ is now known as attila_lendvai
***robin__ is now known as robin
<unmatched-paren>rekado_: in case you don't know yet, i found out that those ghc .hc files are not human-written <unmatched-paren>if you do `ghc --help` and look for information about the `-C` option, you can see: <unmatched-paren>and also, in an example: '# compile Haskell-produced C (.hc) to assembly language:' <unmatched-paren>so although it's not certain, it seems reasonably likely that .hc means 'C produced by GHC' ***ChanServ sets mode: +o rekado_
<rekado_>unmatched-paren: that’s possible, but they look really good for something that might have been generated. <rekado_>I also think that building GHC with ~4000 lines of very readable generated code is many orders of magnitude better than taking a binary of GHC and calling it a day. <stikonas>unmatched-paren: I looked at those files and I don't think they are computer generated <stikonas>yes, .hc files can be produced by ghc, but that does not mean that .hc file is produced by ghc <stikonas>so I think if rekado_ can bootstrap ghc using those files than that's fine <stikonas>yes, commit titles might be better but that's probably an artifact of cvs to git conversion <stikonas>in any case it's still non-trivial to build it <stikonas>I tried with newer GCC and without Hugs and I wasn't able to <stikonas>well, it was probably mostly Hugs... GCC issues seem to be easily patchable <unmatched-paren>stikonas: ah, you're probably right... if we were unsure, we could just ask them if the files are generated :) <rekado_>my current obstacle is -liberty and -bfd <rekado_>I’m trying to build ghc/interpreter (it’s really just a slightly modified hugs), and that is linked with bfd and liberty <rekado_>the linker complains about missing symbols that the bfd objects reference, but that are provided by liberty.