Reply on 6-2-2-branch

[Simon Marlow]

On 21 July 2004 11:28, Serge D. Mechveliani wrote:

> On  July 20
> I have downloaded by  cvs co -r ghc-6-2-branch ghc hslibs libraries
> the version of
>                       CVSROOT :pserver:anoncvs at glass.cse.ogi.edu:/cvs
> and tested it.
> It this for future  ghc-6.2.2 ?
> 
> I have seen only one minor bug.
> Here follow the reports and questions.

Thanks for testing!

> 1. A bug of a negative number of bytes.
> 
>   ghci ...
>   > :set +s
>   > test "log"
>   ...
>   (58.68 secs, -989876700 bytes)
>   >
> 
> Probably this is due to using Int instead of Integer.
> And as I recall, it was said fixed (!?)

I fixed the Haskell side, but not the C side.  Should now be fixed.
 
> 2. It identifies itself as  ghc-6.2.1.

It still is 6.2.1 :-)  I'll announce a release candidate in a few weeks.

> How to distingush it from the official ghc-6.2.1, by what messages?
> Could the CVS versions aiming at some  ghc-XXX,  respond,
> for example, with
>                    " ... GHC, pre-release XXX-pre of July 20, 2004 "
> 
> when called        ghc --version
> ?

Yes, I'd like to do this, but I don't know a good way.  Currently our
nightly build system fixes the version numbers, but not if you get the
code from CVS.  And I don't really want to automatically check-in a
version update once a day, but perhaps that's the only way.

I think the gcc folks manage to do it somehow - anyone know how?  

> 3. `make' produces several warnings about the data fields not
>    initialised:
>   -------------------------------
>   typecheck/TcRnTypes.lhs:235:
>     Warning: Fields of `TopEnv' not initialised: top_nc
>              In the record construction: TopEnv
>                                              {top_mode = mode,
>                                               top_dflags = dflags,
>                                               top_hpt = hpt,
>                                               top_eps = ref,
>                                               top_us = us,
>                                               top_errs = top_errs top}
>              In the definition of `top'':
>                  top' = TopEnv
>   ....
>   -------------------------------

These can be ignored.

Cheers,
	Simon