I gather I should have sent this report to cs.yale.edu: the web page
http://www.haskell.org/hugs/bug-reports.html suggests the address
haskell.org, but I presume as I got no reply that my original mail was lost.
P.S. Why do I get a control stack overflow? Something as "simple" as sort
[1..10000] does this. I had hoped that the control stack would use the
heap, but presumably this isn't so?
-----Original Message-----
From: michael abbott [mailto:[EMAIL PROTECTED]]
Sent: 07 March 1999 15:58
To: '[EMAIL PROTECTED]'
Subject: Problems with Hugs98, 22 Feb 99 version
I've encountered two rather obvious problems with the latest Hugs98 beta
(monad.hs and exts\Channel.lhs won't load without changes) which are easy to
fix, and a strange quirk trying to run Fran (at first I though that Fran
just didn't work at all).
Problems:
1. Can't load Monad.hs!
As far as I can tell, the problem is that the following replacement is
needed from 1.4 to 98: accumulate -> sequence, sequence -> sequence_.
I've made this change to monad.hs, and it seems to load ok.
2. Can't load Graphics\GTest.hs.
This seems to be the same problem, in exts\Channel.lhs; again, changing
sequence -> sequence_ seems to fix things.
3. A rather strange problem trying to run Fran.
I first tried to run Fran\demos\TestMain from the Fran directory (which
fails); changing to the demos directory and trying again then also fails.
However, if I restart Hugs and run TestMain straight away from the demos
directory it seems to work ok. This suggests to me that there's a problem
somewhere in cleaning up on failure or interruption.
To reinforce this impression, I ran TestMain again and type ^C to
interrupt: in this case, I'm left with the Fran animation window until I
close Hugs (I can't close the Fran window itself).
Main>:!cd
C:\Program Files\hugs98-990222-all\lib\fran
Main> main
Let's see something now.
User error: Could not open file "../Media/PatBig.bmp"
Program error: {undefined}
Main> :cd demos
Main> main
Let's see something now.
Program error: {undefined}
Main>:!cd
C:\Program Files\hugs98-990222-all\lib\fran\demos
Main>
When I first got this result, I concluded for a while that Fran just didn't
work (on my machine).
4. The default heap size (100,000 cells) seems remarkably small,
particularly as the heap isn't expandable. Fran won't load at all in this.
A default closer to 1,000,000 would seem more sensible under Win32 (and I've
set mine to 4,000,000 which works fine).
Otherwise, this looks rather promising! When's the full release likely?
Related to that, do you know when GHC with a run-time DLL and without
cygwin32 is likely to come?
Regards,
Michael Abbott
