Re: GHC-5.02/Win32: problems compiling with -prof and -package

2001-10-14 Thread Mike Thomas 

Hi again.

> > Nothing deep going on, the profiling interface files for the 'util'
> > package are simply just missing from the 5.02 distribution.
>
> Sorry, I'll fix that next week in a new InstallShield release (together
> with the other Win32-specific problems reported with 5.02 so far).

Before you do, two other minor points which might be worthy of your regard:

1. Although GHC under Win32 certainly does handle spaces in the installation
directory path, it will not necessarily work when doing "make install" with
the CVS sources of HDirect (and I assume also other utilities in fptools).

My suggestion is that to avoid inconvenience, maybe a warning should be
present in the installation dialogs to help out those of us who are slaves
of CVS fashion.  Perhaps an up-to-date description of the canonical Win32
CVS compilation setup should also be made.

I acknowledge that this issue may be due to my particular build setup:

  - Current Cygwin net release
  - Latest mingw32 tools from sourceforge
  - Latest GHC 5.02 installer from Glasgow
  - set path to firstly scan GHC 5.02 bin directory, then Mingw32 bin
directory, then Cygwin bin directory.
  - configure --build=mingw32 --prefix=c:/Program\ Files/ghc/ghc-5.02

which puts Mingw32 "make.exe" and "gcc.exe" ahead of Cygwin equivalents,
while allowing access to the remainder of the Cygwin tools for the fptools
build.

The end result of  make install in "hdirect" is that the prefix string is
split and makes a directory called "c:/Program".  To avoid this I
reinstalled GHC in a path with no spaces and reconfigured with the
appropriate prefix setting..

2. A bootstrap build of GHC (ie just starting with GHC and no other utility
such as Happy installed) and HDirect from CVS does not go according to plan
in terms of dependencies.  For example, if Happy is missing then the
makefile does not build "happy-inplace" before trying to build other parts
of the CVS source tree which are dependent on Happy (if my memory is
correct, the GHC subdirectory "make boot" or "make" commands fail because
Happy is not built.

I find I often have to go into various sub-directories and type various
combinations of "make boot" and "make lib" before going back to "make".

On the positive side, I have never had a problem building Happy under Win32,
although "make install" does the Unixy thing.

Cheers

Mike Thomas.



___
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: GHC-5.02/Win32: problems compiling with -prof and -package

2001-10-13 Thread Reuben Thomas 

> Nothing deep going on, the profiling interface files for the 'util'
> package are simply just missing from the 5.02 distribution.

Sorry, I'll fix that next week in a new InstallShield release (together
with the other Win32-specific problems reported with 5.02 so far).


___
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: GHC-5.02/Win32: problems compiling with -prof and -package

2001-10-13 Thread Sigbjorn Finne 

Nothing deep going on, the profiling interface files for the 'util'
package are simply just missing from the 5.02 distribution.

--sigbjorn

- Original Message -
From: "P Jensen" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Friday, October 12, 2001 14:00
Subject: GHC-5.02/Win32: problems compiling with -prof and -package


> Hi GHCers --
>
> I'm running GHC 5.02 on Win32 (2000 Professional), and having problems
> compiling with the "-prof" option. There seems to be some sort of conflict
> between the -package option and the -prof option.
>
> For instance, this program:
>
> >module Main where
> >main = putStr "Hello, world.\n"
>
> compiles fine with:
>
>   ghc -prof -auto-all -o t1 t1.lhs
>
> and this program:
>
> >module Main where
> >import GetOpt
> >main = putStr "Hello, world.\n"
>
> compiles fine like this:
>
>   ghc -package util -o t2 t2.lhs
>
> but when compiled like this:
>
>   ghc -prof -auto-all -package util -o t2 t2.lhs
>
> it results in an error:
>
>   t2.lhs:2:
>   failed to load interface for `GetOpt':
>   Could not find interface file for `GetOpt'
>
> My guess is that this is some sort of configuration problem on my end, or
a
> grave misunderstanding of how to invoke the compiler/profiler.
>
> Does anyone have any ideas?
>
> Cheers,
> Paul
>



___
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs