Conor McBride wrote:
Johan Glimming wrote:
I am trying to compile Epigram under Mac OS X with the following problem:

[EMAIL PROTECTED] epigram-20041017_0400]$ make
cd src && make
if [ ! -r . ]; then mkdir .; fi
ghc -package lang -fglasgow-exts -fallow-overlapping-instances
-fallow-undecidable-instances  -O  -odir. -i -i. -hidir.    --make
Main.lhs -optl-s -o ./Epigram-bin
Chasing modules from: Main.lhs
Compiling Utils            ( ./Utils.lhs, ./Utils.o )
/tmp/ghc352.hc:3:17: Stg.h: No such file or directory
/tmp/ghc352.hc:4:20: HsBase.h: No such file or directory
/tmp/ghc352.hc:5:20: HsLang.h: No such file or directory

When it works (please advise how to proceed!)

Your ghc is broken ;-)

Which version of OS X and which ghc package are you using?

I'm using OS X 10.3.5 and the Apple installer package of ghc 6.2.1 from http://www.haskell.org/ghc which is maintained by Wolfgang Thaller. I do notice that this particular flavour of ghc hasn't appeared for 6.2.2 yet, whereas the darwinports version was the first to appear after the initial source and i386/linux binary packages.

I did try the darwinports version, but I couldn't get ghc to build using it. That was some months ago though.


I'd be happy to send you a
Mac binary on a regular basis. In fact, I can give you an account on my
Mac glimming.com so that you can prepare it yourself and use your own
Makefile put command, if you'd like to.

Thanks for the offer: I do think that we ought to have some arrangement
for regular updates of Mac binaries as well as Windows. What tends to
happen at the moment is that I do cvs commit on new source code, and
then something automatic which I don't understand happens at dead of
night on a machine in Durham. Paul looks after the Windows version.
So it's not really my call, other than to encourage Mac people (yourself,
James, Michael, any others?) and Durham people (Paul and Edwin) to figure
out what's practical here.

In terms of releasing stuff I think ghc's approach is quite sane. Anyone can get read-only access to cvs and download the latest source. This could also be provided by nightly source tarballs.


When a significant step forward is made a new version is released as source and with a linux binary. The rest of the people who maintain the other binaries then build and test them and they appear on the download page when they're ready.

I think we will soon outgrow the ability to simply have a handful of cron jobs to package epigram for all platforms, particularly if we want to provide sensible installers for windows and OS X. So why bother, just to save someone 10 minutes of compiling?

For the first time a week or so ago we had a situation where the latest build was worse for writing a particular program than its predecessor. I expect that if there are more people coding in the future then this sort of thing will be more of regular occurrence and a release strategy that is in-sync with the pace of development as opposed to the passage of time would be a better idea.

James



This message has been scanned but we cannot guarantee that it and any
attachments are free from viruses or other damaging content: you are
advised to perform your own checks.  Email communications with the
University of Nottingham may be monitored as permitted by UK legislation.



Reply via email to