March 2001 - Hugs-Bugs

Something wrong with Windows 98 installation??
by John Rood 09 Mar '01

09 Mar '01

Dear sir or madam, I have downloaded your file hugs98-Feb2001.msi. However I can't quite figure out what to do with it. I am running Windows 98 and expected to be able to click on your file and have it set itself up automatically. However, when I click on this, Windows responds by asking me what type of file (i.e. utility) I want to use to open this filename suffix. That is, my Windows 98 doesn't recognize what *.msi is supposed to mean. Thinking this means my machine somehow doesn't know how to do a setup (which I actually know not to be the case in fact, i.e. my machine has successfully set up plenty of other things automatically), I tried to download the "installer" which you have a link to. That is, I clicked on the link in the following sentence of your documentation: In the unlikely event that your machine doesn't already have the Windows Installer engine/runtime installed, runtime redistributables can be found here. [<--link] However, this too was unsuccessful, leading to a missing page at Microsoft. How AM I supposed to set this thing up on my machine? -John Rood

1 0

Minor tweaks
by Sigbjorn Finne 06 Mar '01

06 Mar '01

Attached is a patch to Feb2001 which does the following * flushes stderr when :set -h<size> and :set {+,-}98 is used. As is, the informational messages that's printed on stderr may not appear until later with some terminal & CRT combos. * allows the use of "~/" in :cd commands. ~/ is substituted with whatever the environment variable HOME contains. It doesn't do other forms of tilde expansion, but if there's interest I could always add support for ~login/ i.e., first tweak is a bugfix, the second is a feature tweak. BTW, is there a technical reason why toggling between Haskell98 and Hugs mode can only be done at startup time? --sigbjorn

3 2

RE: problem with hiding?
by Simon Peyton-Jones 05 Mar '01

05 Mar '01

| BTW, the FuncMP source code highlighted that the version of GHC used | to develop FuncMP didn't implement module re-exportation properly - I | had to tweak one of the modules (FMPMpsyntax) quite a bit to have it | work with Hugs98 (and ghc-4.08) Is this a bug report for GHC 4.08? Simon

1 0

problem with hiding?
by Christian Sievers 03 Mar '01

03 Mar '01

Hello, there may be a bug with hiding imports. I tried to load Functional Metapost (http://www.informatik.uni-bonn.de/~ralf/FuncMP.tar.gz) and expected all sorts of problems as it is described to be written in Haskell 1.4, but not this one: Prelude> :l FMPMain [...] ERROR FMPMain.lhs - Entity "space" imported from module "FMPPP" already defined in module "FMPPicture" However, the file FMPMain.lhs contains the line > import FMPPicture hiding (text, empty, space) I cannot see what's going here: simple tests with hiding worked well. I'm using the Feb 2001 version. All the best Christian Sievers

2 2

Re: problem with hiding?
by Sigbjorn Finne 03 Mar '01

03 Mar '01

Johan Nordlander nordland(a)cse.ogi.edu writes: > >> Christian Sievers wrote: >> Hello, >> >> there may be a bug with hiding imports. >> >> I tried to load Functional Metapost >> (http://www.informatik.uni-bonn.de/~ralf/FuncMP.tar.gz) >> and expected all sorts of problems as it is described to be written >> in Haskell 1.4, but not this one: >> >> Prelude> :l FMPMain >> [...] >> ERROR FMPMain.lhs - Entity "space" imported from module "FMPPP" >> already defined in module "FMPPicture" >> >> However, the file FMPMain.lhs contains the line >> > import FMPPicture hiding (text,empty,space) >> >> I cannot see what's goin here: simple tests with hiding worked well. >> I'm using the Feb 2001 version. > > > The reason this example doesn't work is because of the module > FMPMpsyntax, which imports FMPPicture selectively, and then reexports > FMPPicture in the module header. Hugs doesn't implement this correctly, > it reexports all identifiers of every module mentioned in its export list. I've been bitten by this infelicity numerous times too, and started a while ago to modify the sources so as to fix it, but never got around to completing the job. The above report caused me to finally do something about it, so attached is a patch which fixes this problem for Hugs98 (wrt Feb 2001 sources.) Hopefully the patch is in a decent enough shape for it to be included & that it will prove useful to others. BTW, the FuncMP source code highlighted that the version of GHC used to develop FuncMP didn't implement module re-exportation properly - I had to tweak one of the modules (FMPMpsyntax) quite a bit to have it work with Hugs98 (and ghc-4.08) --sigbjorn

1 0

Feb2001 hugs98 on MacOS
by Richard A. O'Keefe 03 Mar '01

03 Mar '01

The February 2001 release of Hugs 98 unpacked fine and seems to run fine on my PowerMac 7600/120 running Mac OS 8.6 HOWEVER, the characters in the console window are teeny tiny wee characters indeed. What are they, 8 point? I could live with 10 point, but would personally rather have 12 point. So what do I do to change the size of the text in this window? I poked around in the sources, and discovered that the window is a SIOUX console. It appears that Hugs does not set any SIOUX options to control the text size, but I may have missed something. I wanted to try setting SIOUXSettings.fontsize = 12; but unfortunately I have CW Pro 2, not CW Pro 5, so cannot use the XML project file provided, and no instructions for using CW Pro 2 were included. I would strongly recommend adding a "-fontsize" command line option that can be put in the Hugs Preferences file. Come to think of it, a -fontface option so that I can get 'courier' instead of 'monaco' would be nice too.

4 11

strange bug
by Andreas Marth 03 Mar '01

03 Mar '01

Hi again! I cut the example a bit further down. Read the comments at the end of the file. --------------------------------------------------------------------------------------------------------------- >import Array > >data CM a = ZM | M (Array (Int,Int) a) deriving (Show,Eq) > >data RD = NB !(Double,Int) > >main = do let arr = foo (NB (1.0,1)) > putStr ("arr==arrS "++show (arr==arrS)++"\n") > putStr ("arrS==arr "++show (arrS==arr)++"\n") > putStr ("bnds arr arrS "++show ((bounds arr)==(bounds arrS))++"\n") > putStr ("bnds +id arr arrS "++show (((bounds.id) arr)==((bounds) arrS))++"\n") > putStr ("id +bnds arr arrS "++show (((id.bounds) arr)==((bounds) arrS))++"\n") > >foo :: RD -> Array (Int,Int) (CM Double) >foo rd = case rd of > NB (f,p) -> h where h=hCMt [(p,listArray ((1,1),(1,1)) [f])] > where > --h0CMt :: Array (Int, Int) (CM Double) > h0CMt = accumArray (\a _-> a) ZM ((1,1),(1,2)) [] > hCMt po_arL = h0CMt // (map (\(po,ar) -> ((1,po),M ar)) po_arL) > >arrS :: Array (Int,Int) (CM Double) >arrS = listArray ((1,1),(1,2)) [M (listArray ((1,1),(1,1)) [1.0]),ZM] >-- This is the array that foo (NB (1.0,1)) shows in Hugs. >-- But (foo (NB (1.0,1)))==arrS is False. >-- All in main is only to show the strange behaviour. >-- If I write NB (f,p) -> hCMt [(p,listArray ((1,1),(1,1)) [f])] instead of line 16 >-- the bug disappears. That is also the reason why I have to keep the data declaration RD. >-- If I put the type signature of line 18 in scope the bug also disappears. >-- If I write >-- hCMt po_arL = (accumArray (\a _-> a) ZM ((1,1),(1,2)) []) // (map (\(po,ar) -> ((1,po),M ar)) po_arL) >-- instead of line 19 and 20 it also vanishes. >-- So I thing now it is the shortest example that I can find. (Apart from deleting main.)

1 0

strange bug
by andreas.marth＠daimlerchrysler.com 02 Mar '01

02 Mar '01

Hallo everybody! The file bug1.hs includes an example where acording to main sT==sXXP is True, but sXXP==sT is False. The file bug2.hs doesn't have the problem after inlinige one where clause. The file bug3.hs also doesn't have the problem after adding a type signature to the function "h0CMt". So what exactly is the problem here? Thanks, Andreas

2 1

Re: possible bug?
by Sigbjorn Finne 02 Mar '01

02 Mar '01

Jonathon Bell jbell(a)mathsoft.com writes: > > i'm tinkering with the Feb 2001 release of hugs, and came across the > following problem while attempting to create my own prelude: > > module MyPrelude where > > import Prelude((:)) > > hd(x:xs) = x > ... > > The compiler complains of a syntax error occuring in the import declaration: > (unexpected symbol ":"). I may well be mistaken since i am somewhat new to > haskell programming, but it seems to me that i ought to be able to > explicitly import the list type constructor just like any other identifier No, the cons operator has special status in H98 (in earlier version of Haskell, this wasn't the case); it is considered to be part of the syntax for lists, just like [] and [x1,x2] is. See the discussion of reservedops in Section 2.4 of the report. hth --sigbjorn

1 0

possible bug?
by Jonathon Bell 02 Mar '01

02 Mar '01

i'm tinkering with the Feb 2001 release of hugs, and came across the following problem while attempting to create my own prelude: module MyPrelude where import Prelude((:)) hd(x:xs) = x ... The compiler complains of a syntax error occuring in the import declaration: (unexpected symbol ":"). I may well be mistaken since i am somewhat new to haskell programming, but it seems to me that i ought to be able to explicitly import the list type constructor just like any other identifier. Thanks for a doing a great job supporting hugs, and keep up the good work... ________________________________ Jonathon Bell jbell(a)mathsoft.com MathSoft, Inc. www.mathsoft.com 101 Main St, Cambridge, MA 02142 (617) 577-1017 x745

3 3