Hi
Yhc has basic support for recursively dependant modules - you can
write your own .hi file - I've openned a bug for something better, but
I think its dependant on having a better type checker, so we can type
check modules in parallel.
Thanks
Neil
On 8/18/06, Isaac
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Neil Mitchell wrote:
Hi,
I have filed a bug on this: http://code.google.com/p/yhc/issues/detail?id=15
Does this actually conflict with anything you are trying to do for real? (if so we'll bump the priority a bit).
No, my only current "for-real" thing still relies on recursively dependent modules (and found a GHC bug where it choked on them too, so I'm not making much progress on it currently... along with having other things to do, (like find obscure bugs in Yhc :))
In fact this is in my Main.lhs, and I don't believe it would be affected by these bugs:
Avoid non-explicit use of floating-point numbers -- they're inexact, platform/machine-dependent, and mostly unnecessary. Of course relying on defaults at all will be generally avoided. In fact, since these defaults are confusing and rare, disable them altogether. In particular, they are the only thing besides the also-bad monomorphism restriction that make module boundaries significant -- so make sure to disable defaults in every module, like this:
default ( ) -- defaults are bad, see Main
BTW it's an attempt to reimplement RoboWar in Haskell, and the code is coming out much cleaner than the original (which was written in C only for Mac pre-OSX), but after about a week of working on it earlier this summer I stopped when I could only compile it with GHC 6.5.something (though the GHC bug will be fixed in 6.4.3, and if I felt like it I might be able to add workarounds to the .hs-boot files...)
Isaac -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFE5aUDHgcxvIWYTTURAmqDAJ43g68c9DiOMfKjJdsdcPtldc/Q0gCfV7CO SrTg/EeMGvLnvtZP9vd3Ass= =jjBp -----END PGP SIGNATURE-----