[Cc'ing hugs-users]
Thanks, this is a long-standing and known feature of Hugs' interact
implementation, so before possibly bringing it into line with what
Haskell 98 tells us how `interact' ought to behave regarding echoing
of input, it would be prudent to first check whether anyone depends
on the current behaviour. In particular, are there 'important' resources
(e.g., textbooks and tutorials) based around Hugs98 that depend on
the current non-echoing behaviour? Or are there other weighty
arguments in favour of maintaining status quo?
--sigbjorn
----- Original Message -----
From: "Malcolm Wallace"
There is a bug in Dec2001 Hugs (and earlier releases too) which causes the Prelude function `interact' to behave unexpectedly - the user's input is not echoed to the screen as she types it. A patch is attached which solves the problem.
Regards, Malcolm
--- src/iomonad.c Tue Dec 11 00:54:51 2001 +++ src/iomonad.c.patched Tue Dec 18 10:53:47 2001 @@ -844,6 +844,7 @@ HandleArg(h,1); if (handles[h].hmode&HSEMICLOSED) { /* read requires semi-closed handle*/ Int c = (h==HSTDIN ? readTerminalChar() : getc(handles[h].hfp)); + if (h==HSTDIN) putchar(c); if (c!=EOF && c>=0 && c
_______________________________________________ Hugs-Bugs mailing list Hugs-Bugs@haskell.org http://www.haskell.org/mailman/listinfo/hugs-bugs
There is a bug in Dec2001 Hugs (and earlier releases too) which causes the Prelude function `interact' to behave unexpectedly - the user's input is not echoed to the screen as she types it. A patch is attached which solves the problem.
Thanks, this is a long-standing and known feature of Hugs' interact implementation, so before possibly bringing it into line with what Haskell 98 tells us how `interact' ought to behave regarding echoing of input, it would be prudent to first check whether anyone depends on the current behaviour. In particular, are there 'important' resources (e.g., textbooks and tutorials) based around Hugs98 that depend on the current non-echoing behaviour?
Bird (2nd ed) makes use of the non-echoing interact, and describes another, programmed, solution to restore echoing, all on a single page (pp.230). Regards, Malcolm
"Malcolm Wallace"
Thanks, this is a long-standing and known feature of Hugs' interact implementation, [...] are there 'important' resources (e.g., textbooks and tutorials) based around Hugs98 that depend on the current non-echoing behaviour?
Bird (2nd ed) makes use of the non-echoing interact, and describes another, programmed, solution to restore echoing, all on a single page (pp.230).
ok, thanks - i'll look this up once back from the holidays. As one half of the new Hugs maintainer duo (Galois Connections is taking over maintenance of Hugs -- Jeff Lewis being the other half), I think it is important we pay special attention to the needs of book authors and teachers of Haskell, as we go forward. There are a number of changes I'd like to see Hugs adopt: better Haskell98 compliance (e.g., clean up the Prelude, adopt your change to Prelude.interact), move over to use the new hierachic library structure etc. These changes are likely to have a number of minor upsetting effects on existing code, so finding ways to minimise or avoid these alltogether needs to be considered first. My current thinking is to try stabilise the current sources until we end up with a release that proves to be stable, and then probably fork the development tree. If anyone using Hugs for teaching or rely on it for a textbook have feedback on how Hugs could better serve their needs, I'd be really interested to hear from them. --sigbjorn
participants (2)
-
Malcolm Wallace -
Sigbjorn Finne