[EMAIL PROTECTED] writes:
> Hi,
>
> These are version 2.02 bugs; I've run into the problem building
> 2.04-pl2 using a 2.02 binary distribution described by Mariano Suarez
> Alvarez. It seems like the Makefiles aren't using mkdependHS at all,
> but are just trying to compile the *.hs files in alphabetical order
> (starting with parser/U_binding.hs) But I don't know how to work
> around this...
>
Hi,
thanks for the report, sorry about the long delay. I'm guessing that
`make boot' didn't succeed in making the dependencies in ghc/compiler,
could you check the output of `make boot' to see?
> First bug:
> This compiles and links using the "in-place" ghc-2.02 alpha-dec-osf1
> binary distribution, but fails using the "in-place" ghc-2.02
> sparc-sun-solaris2 binary distribution, with a linker error. Comments
> on speeding the program up would also be appreciated. The source code,
> and a script of the compile attempt, are at
>
This is likely to be a Solaris swap space problem (linker reports
"..cannot msync file; errno=49"); add/free up some more and you should
be OK..
> http://www.cs.umd.edu/~burdick/haskell/bug.html
>
> The second bug report is at the end of this message. It seems to be a
> problem with unboxed types. Again, I was using the ghc-2.02 solaris
> "in-place" binary distribution.
>
...
>
> n = 3# +# 4#
>
...
>
> panic! (the `impossible' happened):
> mkPrimReturnCode: Upd
>
> Please report it as a compiler bug to [EMAIL PROTECTED]
>
...
Thanks, this toplevel unboxed value should have been magically lifted,
but was not. Fixed in the next release.
--Sigbjorn
