September 2003 - Hugs-Bugs

parsing strings
by Iavor Diatchki 08 Oct '03

08 Oct '03

hello, there seems to be a problem when parsing multi-line strings: in the first string there is no space after the \, while in the second there is. i tried this on linux, with the july version of hugs. > probelm = "hello \ > \world" > noproblem = "hello \ > \world" the error i get is: "Illegal character escape sequence "\w" bye iavor -- ================================================== | Iavor S. Diatchki, Ph.D. student | | Department of Computer Science and Engineering | | School of OGI at OHSU | | http://www.cse.ogi.edu/~diatchki | ==================================================

2 1

problems with export declarations in hugs nov 2002
by Andrew Frank 30 Sep '03

30 Sep '03

i am moving (under windows) from hugs nov 99 to hugs nov 02. i observe the following difference: i understand that for classes and data types, an export of the form X (..) exports the class/data type with all the methods/constructors in scope (report page 64). This seems not to work module A (Eq (..), Bool (..)) where -- produces an error in module b (not under hugs nov99) --module A (Eq ((==)), Bool (False)) where -- works (but should not be necessary) import Prelude (Eq (..), Bool (..)) f = 'a' == 'b' --------------------------------------------------- module B where import qualified Prelude import A t = 'a' == 'a' f = False ------------------------------------------------------ what do i misunderstand? (by the way, this seems only to occur when the class is shadowing a class in the prelude) thank you andrew Andrew U. Frank Geoinformation E127 phone: +43 1 588 01 12710 TU Vienna secr. +43 1 588 01 12700 Gusshausstrasse 27-29 fax +43 1 588 01 12799 A-1040 Vienna Austria cellular phone +43 676 41925 72 http://www.geoinfo.tuwien.ac.at/persons/frank/frank.html

2 1

small hugs98-Sep2003 build problem on FreeBSD 4.8
by Reinhard Speyerer 28 Sep '03

28 Sep '03

Trying to install hugs98-Sep2003 on FreeBSD 4.8 produced the following error: $ ./configure ... checking for function_dlopen... yes checking for function_shl_load... no checking for library_dld... no ... $ cd .. $ make ... env HUGSDIR='../hugsdir' HUGSFLAGS='-W' unix/BuildFFI Compiling ../../hugsdir/libraries/Foreign/C/Error.hs Compiling ../../hugsdir/libraries/Foreign/Marshal/Alloc.hs Compiling ../../hugsdir/libraries/Foreign/Marshal/Utils.hs Compiling ../../hugsdir/libraries/System/Posix/Internals.hs Compiling ../../hugsdir/libraries/Text/Regex/Posix.hs Compiling ../../hugsdir/libraries/Network/BSD.hs Compiling ../../hugsdir/libraries/Network/Socket.hs Compiling ../../hugsdir/libraries/System/Posix/Directory.hs /usr/libexec/elf/ld: cannot find -ldl runhugs: Error occurred ERROR "../../hugsdir/libraries/System/Posix/Directory.hs" - Error while running compilation command 'gcc -g -shared -fPIC -D__HUGS__ "-I../hugsdir/include" "../../hugsdir/libraries/System/Posix/Directory.c" -o "../../hugsdir/libraries/System/Posix/Directory.so" -I../../hugsdir/libraries/System/Posix -I../../hugsdir/libraries -I../../hugsdir/include -I/home/rls/hugs98-Sep2003/fptools/libraries/unix/include -ldl' The problem seems to be caused by the fact that FreeBSD (and possibly other BSD-derived systems) have dlopen() etc. in libc, there is no libdl. configure #defines HAVE_LIBDL to 1 if either dlopen() etc. are found in libc or when linking with -ldl. fptools/libraries/unix/package.conf.in however uses #ifdef HAVE_LIBDL to check whether -ldl should be added to extra_libraries. After manually deleting the #ifdef HAVE_LIBDL , "dl" #endif block from package.conf.in make completed successfully and make LibStatus completed without producing errors. Regards, Reinhard

2 1

non-bug report
by Doug McIlroy 24 Sep '03

24 Sep '03

I just installed hugs98-Nov2002 on Linux. Absolutely smooth--a great improvement over hugs 1.4, which needed considerable fiddling back in 2001. yours delightedly, Doug McIlroy

1 0

Re: Hugs and osX 10.3?
by Alastair Reid 23 Sep '03

23 Sep '03

[moved from hugs-authors to hugs-bugs (which is where most development-related discussion takes place] > I use Mac OS X, and Apple plans a new release of 10.3 > very soon. Is Hugs still in development (I see the > last version was Nov 2002), and is there a OS X 10.3 > release planned? Hugs is still very much in development and is available for MacOS X 10.2. A new source release of Hugs is due in the next week or so. Binary releases for popular platforms are usually donated by volunteers within a few days of a source release. -- Alastair Reid www.haskell-consulting.com

2 1

qualified import of prelude
by Andrew Frank 22 Sep '03

22 Sep '03

i seem to have a problem with a qualified import of the prelude in hugs nov03, haskell extensions, under windows: Reading file "C:\gisTheoryCodeSept03\lib\Language\Value240.hs": ERROR "C:\gisTheoryCodeSept03\lib\Language\Value240.hs":207 - Ambiguous variable occurrence "quotRem" *** Could refer to: Language.Value240.quotRem Hugs.Prelude.quotRem the relevant code is: instance Ring Float where quotRem a b = (toFloat i, toFloat (b * f)) where ab = a/b i::Int (i,f) = Prelude.properFraction ab rem a b = f where (i,f) = quotRem a b --here is line 240 with the error (*) = (Prelude.*) quot a b = i where (i,f) = quotRem a b div = error "div not implemented for float" mod a b = error "mod not implemented for float" i had a similar error (redefining a class Integral and a complaint in the export list of ambiguit between the local definition and the qualified import). it seems that qualified import of the prelude is not properly handled? unfortunately this is not a small file and i did not manage to isolate the problem quickly in a small example. can you see what is wrong here? the file value240.hs is attached. thank you for your attention andrew Andrew U. Frank Geoinformation E127 phone: +43 1 588 01 12710 TU Vienna secr. +43 1 588 01 12700 Gusshausstrasse 27-29 fax +43 1 588 01 12799 A-1040 Vienna Austria cellular phone +43 676 41925 72 http://www.geoinfo.tuwien.ac.at/persons/frank/frank.html

2 1

-W option
by Paul Hudak 19 Sep '03

19 Sep '03

This isn't really a bug report, but I wasn't sure where else to ask the following question. Hugs used to handle all of my SOE code with no problem. Now, however, I get the error message: ERROR "C:\Program Files\Hugs98\lib\exts\MarshalUtils.hs":2 - Syntax error in module definition (unexpected keyword "with "; possible cause: -W option not used.) Indeed, when I invoke Hugs -W from a command line, everything works fine. So the problem is apparently because Hugs used to allow this language extension as a default, and now doesn't. I would hate to tell everyone using SOE that they have to invoke Hugs from a command prompt, so I'm looking for an alternative suggestion if you have any. Of course, I could rewrite the code, but it's actually Alastair's code, and I'm not sure how big of an effort it is... Thanks, -Paul

5 8

illegal export of lone data constructor
by Andrew Frank 17 Sep '03

17 Sep '03

i am moving fom the nov 99 to the nov 02 hugs.exe (under windows) when recompiling old files i find two problems which i do not understand: i want to export (:) from the prelude because i do not use the standard prelude usually. module XX ( (:), ....) does not work (it did before). now i get the error "illegal export of lone data constructor" i do not see any hint in the report (nov 99) that (:) cannot be exported. (it may be unusual, but not illegal?) the goal is to export certain functions from the prelude (especially (:), which cannot be redefined) but not all of it. how can this be achieved? (minor point: my report says that the prelude module is always available as qualified import (p. 71) the hugs now (2002) seem to require an explicite import qualified (after a import Prelude hiding (stuff)) ) i would appreciate an answer on the first question (export of (:)) andrew Andrew U. Frank Geoinformation E127 phone: +43 1 588 01 12710 TU Vienna secr. +43 1 588 01 12700 Gusshausstrasse 27-29 fax +43 1 588 01 12799 A-1040 Vienna Austria cellular phone +43 676 41925 72 http://www.geoinfo.tuwien.ac.at/persons/frank/frank.html

4 3

winhugs
by Richard Lawlor 16 Sep '03

16 Sep '03

1 0

type synonym or fun dep problem
by Iavor Diatchki 16 Sep '03

16 Sep '03

apologies if this gets to the list twice. --- hello, there seems to be something wrong with the implementation of functional dependencies, or perhaps the expansion of type synonyms. i am using hugs (september 2003). the program: > module Test where > > class C a b | a -> b where > mk :: IO a > eq :: a -> a -> IO Bool > > type F = IO > > instance C a x => C [a] (F a) > > test ys = do x <- mk > xs <- mk > eq ys (x:xs) causes the error: ERROR "test.lhs":18 - Constraints are not consistent with functional dependency *** Constraint : C [a] (IO a a) *** And constraint : C [a] (F a) *** For class : C a b *** Break dependency : a -> b if 'type F = IO' is replaced with 'type F a = IO a' one gets *** Constraint : C [a] (IO [a]) if i inline the definition of 'F' things seem to work. if i remove the assumption on the instance: *** Constraint : C [a] (b a) if i remove the assumption on the instance, and eta-expand the type synonym: *** Constraint : C [a] (IO (IO a)) bye iavor -- ================================================== | Iavor S. Diatchki, Ph.D. student | | Department of Computer Science and Engineering | | School of OGI at OHSU | | http://www.cse.ogi.edu/~diatchki | ==================================================

1 0