- Glasgow-haskell-users

RE: GHC Alpha port?
by Simon Marlow 16 Jul '01

16 Jul '01

> On 2001-07-11T10:28:33+0100, Simon Marlow wrote: > > > I wonder if it matters whether PLATFORM_CC_OPTS is set to > -static in > > > mk/bootstrap.mk? The comments there say that the flags > "should match > > > the list in machdepCCOpts in ghc/compiler/DriverFlags.hs", and > > > DriverFlags.hs does set -static for alpha. > > > > I'm not sure whether -static is needed; try it without and see what > > happens. > > I have no idea why, but... > > puffin:~/u/ghc-port/alpha/ghc/compiler$ ./ghc-inplace > Alloc Collect Live GC GC TOT TOT > Page Flts > bytes bytes bytes user elap user elap > ASSERTION FAILED: file GC.c, line 1347 > IOT/Abort trap (core dumped) > > The assertion in question is: > > /* make sure the info pointer is into text space */ > ASSERT(q && (LOOKS_LIKE_GHC_INFO(GET_INFO(q)) > || IS_HUGS_CONSTR_INFO(GET_INFO(q)))); > > It seems that the GC code is sensitive to the layout of the virtual > memory address space. In particular, I had to change HEAP_BASE from > 0x50000000 to 0x200000000L in MBlock.h to get GC to work even with > -static. So it doesn't work without -static? A HEAP_BASE change is not unexpected, it all depends where the system puts its shared libraries. > I assume it is supposed to be able to find these .hi files? The > hc-build script calls for > > ./configure > $MAKE -C ghc/utils clean boot > # ... > > (the "./configure" in which I interpret as um, shorthand for > > GHC=<the ghc-inplace I just built from the .hc files> \ > HappyCmd=<a fake happy that simulates happy -v output> \ > ./configure > > ), but making boot in ghc/utils involves "ghc-inplace -M", which > fails: > > $ .../ghc-inplace -M -optdep-f -optdep.depend -osuf o > -cpp -DPKG_TOOL -DWANT_PRETTY -package lang -package util > -package text Main.hs Package.hs > Main.hs: can't find one of the following: `IO.hi' `IO.hs' `IO.lhs' > > So... perhaps it would be obvious to you what I am missing? The hc-build script is a little broken. What you actually need to do is: - Take the .hs files generated from the .hsc files from the host to the target (these should be part of the .hc file bundle). - Build in ghc/lib/std and hslibs before ghc/utils, - Then build in ghc/utils using the newly built libraries. Cheers, Simon

1 0

RE: GHC Alpha port?
by Simon Marlow 16 Jul '01

16 Jul '01

> > The relevant makefile code is in fptools/mk/bootstrap.mk - I'm sure > > you'll be able to modify it to do the right thing. When > you've got it > > working we'll think about how to set up the bootstrapping system so > > mangling is optional. > > I wonder if it matters whether PLATFORM_CC_OPTS is set to -static in > mk/bootstrap.mk? The comments there say that the flags "should match > the list in machdepCCOpts in ghc/compiler/DriverFlags.hs", and > DriverFlags.hs does set -static for alpha. I'm not sure whether -static is needed; try it without and see what happens. > (By the way, are *_hsc.[hc] files considered part of the ".hc > distribution"? I needed them for compiling, say, Directory.hc.) > > Anyway, the current hurdle is that > ghc/lib/std/PrelInt.hc doesn't compile > because int64ToIntegerzh_fast is not declared > because ghc/includes/PrimOps.h did not declare it > because SUPPORT_LONG_LONGS is not defined in ghc/includes/StgTypes.h > because SUPPORT_LONG_LONGS is defined in ghc/includes/StgTypes.h > if and only if HAVE_LONG_LONG && SIZEOF_VOID_P < 8, > and SIZEOF_VOID_P is 8. > > I'm not sure what the right fix is here. Why does StgTypes.h not > define SUPPORT_LONG_LONGS if SIZEOF_VOID_P >= 8, anyway? I think it's right not to define SUPPORT_LONG_LONGS - as far as I can tell, that symbol means "we need the long long type for StgWord64/StgInt64 on this architecture", which isn't true on Alpha. On a 64-bit machine, int64ToIntegerzh_fast and word64ToIntegerzh_fast are identical to intToIntegerzh_fast and wordToIntegerzh_fast respectively (as long as GMP is using a 64-bit limb type, which fortunately it does). The folllowing (untested) patch should be along the right lines... Cheers, Simon *** PrimOps.hc 2001/07/06 14:11:38 1.79 --- PrimOps.hc 2001/07/11 09:24:29 *************** *** 594,601 **** FE_ } ! #endif /* HAVE_LONG_LONG */ /* ToDo: this is shockingly inefficient */ --- 594,616 ---- FE_ } + #elif SIZEOF_VOID_P == 8 ! FN_(word64ToIntegerzh_fast) ! { ! FB_ ! JMP_(wordToIntegerzh_fast); ! FE_ ! } ! ! FN_(int64ToIntegerzh_fast) ! { ! FB_ ! JMP_(intToIntegerzh_fast); ! FE_ ! } ! ! #endif /* SUPPORT_LONG_LONGS || SIZEOF_VOID_P == 8 */ /* ToDo: this is shockingly inefficient */

2 2

fromInteger and friends
by Simon Peyton-Jones 13 Jul '01

13 Jul '01

Folks On my bicycle this morning I wondered whether I'd ever advertised the following feature of GHC, which various people (notably Sergey I think) have asked for. In Haskell you can *almost* get all your own arithmetic, by not importing the Prelude, and instead importing (say) MyPrelude. But that doesn't take account of literals: when you say "1" in Haskell you get (Prelude.fromInteger (1::Integer)) fromInteger doesn't need to be in scope; you don't even have to import the Prelude; but you get the fromInteger from the Prelude regardless. In the current GHC 5, if you give the flag -fno-implicit-prelude then a) the prelude is not imported by default and (more important) b) instead of PrelNum.fromInteger (1:Integer) you get simply (fromInteger (1::Integer)) That is, the fromInteger is picked up from whatever is in scope. This rebinding of implicitly mentioned identifiers applies to: * fromInteger and fromRational used in literals * negate, used when you say "- x" * subtraction, used when you have an n+k pattern This should make it possible to completely decouple arithmetic from the Prelude. Booleans, lists, tuples etc are still all bound to their Prelude versions -- it's harder to uncouple them, and there is less demand. This feature is documented, but I don't know whether anyone has ever tried it out, perhaps because I never told anyone that it existed. This message is to rectify that omission. I think there's a bug in the 5.00 implementation that means it'll only work well if you define fromInteger etc and then import then; it won't work properly if you define fromInteger etc in the same module as the literals that are meant to bind to them. This'll be fixed in the next release, but meanwhile shouldn't be much of a bother. Let me know if you try it out. Simon

1 0

Install Shield Version of GHC 5.0 for Win?
by Christian Lescher 13 Jul '01

13 Jul '01

Is there an InstallShield Version of GHC 5.0 for Windows in preparation? Will it be available soon? AFAIK, GHC 4.08.2 is still the latest binary distribution for Win. Christian

2 1

RE: Install Shield Version of GHC 5.0 for Win?
by Simon Peyton-Jones 13 Jul '01

13 Jul '01

Yes! We do now have an installshield for Win32, including ghci, that has the new property that a) you don't need to install anything else b) even if you have installed other things (notably cygwin), ghc won't be confused by their presence Today we will put out a "thrill-seekers release" for Win32. By which I mean that it'll be a release of the HEAD of the tree, which hasn't been through the usual stabilisation process before release, so some dark corners may not work as advertised, and we won't advise gettting rid of your trusty 4.08 until we put out a full 5.02. Simon -----Original Message----- From: Christian Lescher [mailto:christian@lescher.de] Sent: 12 July 2001 20:57 To: glasgow-haskell-users(a)haskell.org Subject: Install Shield Version of GHC 5.0 for Win? Is there an InstallShield Version of GHC 5.0 for Windows in preparation? Will it be available soon? AFAIK, GHC 4.08.2 is still the latest binary distribution for Win. Christian

1 0

RE: Problem with .ghci (fwd)
by Simon Marlow 12 Jul '01

12 Jul '01

> Quite. I don't think there is a solution that's going to work > everywhere > and the current version is simple and effective for most > people. I fixed > my InteractiveUI.hs to ignore g+w if the group that owns the > file is my > group, but that's completely wrong at sites which have some users in > shared groups. I guess there's no reliable general test for a > UPG except > with local knowledge. There's certainly no fix for the shared > directory > situation that isn't also going to let dubious .ghci files through > elsewhere. > > Unfortunately I can't think of any better option than to > leave it as it > is with a note in the manual indicating the default > permissions-checking > behaviour is and where to look in the source to change it if > necessary. To subvert GHCi's paranoia, just put this in your $HOME/.ghci: :def source IO.readFile :source ./.ghci (for extra marks, you could make it check that `pwd` /= $HOME, that you hadn't already loaded ./.ghci). Cheers, Simon PS. quite embarrassingly, it appears that GHCi just goes ahead and reads $HOME/.ghci even if the permissions are unsafe. Time for another release methinks, before GHCi gets its first CERT advisory :)

2 1

Re: Trying to build HDIRECT with ghc-5.00.2
by Sigbjorn Finne 11 Jul '01

11 Jul '01

Adrian Hey ahey(a)iee.org writes: > *** ghc Newbie Alert ** > I suppose the first question i should ask is will the current version > (hdirect 0.17) work with ghc-5.00.2 or am I wasting my time? > I've found I have to put '-package lang' in the ghc options to get > the installation to to work at all, but it stops eventually complaining about > unrecognised flag -K2m (obsolete presumably). Is there a ghc-5.00.2 > alternative? At this point I've given up because I don't really understand > what I'm doing or what are appropriate settings for all those > xyz_HC_OPTS in the makefiles or what else might need changing. > > Any advice? hi, the CVS repository version of HDirect is mostly(*) compatible with ghc-5.00.2, so I'd suggest giving it a go instead of 0.17. The repository hangs off the fptools/ one in fptools/hdirect. --sigbjorn (*) - ghc-5.00.2 give Int.sizeofInt{8,...} the type Int32; HDirect generated sources assume Word32 (as it should be). => you may need to tweak the generated code in a select few places to make ghc-5.00.2 happy.

1 0

RE: Problem with .ghci (fwd)
by Simon Peyton-Jones 11 Jul '01

11 Jul '01

| explicitly tell ghci that they're okay! Hand-holding w.r.t. | 'insecure' file permissions has a nasty habit of becoming a | nuisance in unusual cases the original authors hadn't thought of. :-( Constructive suggestions for how to improve are welcome. What we are trying to avoid is obvious trojan horses, where X can persuade Y's ghci to do rm *.*. This is bad. Simon

3 2

RE: Regexps
by Simon Marlow 11 Jul '01

11 Jul '01

> ["1,0,0,0","","","","","","","","","","","","","","","","","", > "","","","","","","","","","",""] > > ie the right answer, and 28 other empty answers. > > The documentation says > > "Note that there may be more components in the returned list than were > in the pattern, and that any component which either matched the empty > string or wasn't matched at all (because it was part of an > optional part > of the pattern), will be empty." > > Does this mean I had 28 matches with the empty string? No, it means there are more components in the returned list than in your original pattern, and you should ignore the extra ones. This is because GNU regex doesn't tell us how many components are actually returned, as I recall. > Why are they > returned? Will I always have to filter all these null strings out of > every result? Will the result I want always be first? Yes, no(*), and yes. (*) don't filter out the null strings, just strip the extra ones from the tail of the returned list. Remember that some of the components may legitimately be empty, for example if one of the components in your pattern was optional (eg. "a(bc)?d"). Cheers, Simon

1 0

Regexps
by Amanda Clare 11 Jul '01

11 Jul '01

Hi, I'd like to use regular expressions in my code, so I looked at the ghc documentation on this library and it doesn't say much. RegexString looks great but I don't know how to use it. Some examples would be greatly appreciated! Regex itself has even less documentation. From RegexString: "mkRegex makes a regular expression with the default options (multi-line, case-sensitive), whereas mkRegexWithOpts takes the values for these options as parameteres. The syntax of regular expressions is otherwise that of Perl." Then what's wrong with let exp = mkRegex "^class\(\[(.*)\]," in This would be the Perl syntax for the regexp. It says ghc -package text -c -o RegT.o RegT.hs RegT.hs:13: error in character literal I checked and it doesn't like the escaped ( [ and ]. So how do I escape them? The unescaped inner round brackets mean that I want to collect the result of this match, whereas the escaped brackets mean that these are actual characters I want to match. My text to match looks like this: class([1,0,0,0],"METABOLISM"). class([1,1,0,0],"amino-acid metabolism"). class([1,1,1,0],"amino-acid biosynthesis"). Amanda -- Amanda Clare http://users.aber.ac.uk/ajc99/ Tel: +44 (0)1970 621922 Fax: +44 (0)1970 622455 Dept. of Computer Science, University of Wales, Aberystwyth, SY23 3DB

2 2