[OS X TeX] Re: Xindy?

[Joachim Schrod]

Peter Dyballa wrote:

First, an important point. You wrote:

> That these steps 
> cannot always be reproduced makes sure that this software is still 
> experimental and not well adapted to PPC Mac OS X.

This might be a misconception: xindy is not adapted at all to Mac OS X, be it on 
PPC or on Intel. It *is* experimental on a Mac, that's definitivly a true 
statement: The configuration is for Unix, and one user (Maarten) reported that 
it compiled on his Mac when one turns off readline. In this forum, some other 
users reported that they can compile it as well, but many others have problems. 
What I'm trying to do here is to collect information how to adapt configuration 
and compilation to Mac OS X.

I'm sorry to say that the log excerpts that you posted don't give me any clue 
where the problem could be. (As I wrote, I don't have a Mac and cannot try to 
reproduce any of your problems.)

That said, I can comment on a few things.

1) configure call:

  >     ./configure LDFLAGS="-dead_strip -L/usr/local/lib -L/sw/lib/ncurses
> -L/sw/lib -L/usr/X11R6/lib" CPPFLAGS="-I/usr/local/include -I/sw/include 
> -I/usr/X11R6/include" CFLAGS="-ggdb -O -pipe -faltivec -maltivec 
> -mabi=altivec -mcpu=7450 -no-cpp-precomp -fomit-frame-pointer 
> -foptimize-register-move -fcprop-registers -frename-registers 
> -freorder-blocks -fpeephole -mpowerpc-gfxopt -mpowerpc-gpopt"

I don't know what you want to achieve, but this is a sure way to make it fail; 
all of these -f options are bound to cause havoc during compilation. You 
influence low-level register allocation and stack-frame layout decisions; you 
should only do this when you're very sure that you need this and know exactly 
what they will cause in the application.

Has configure problems in detecting your correct platform? Is that the reason 
for those -m options?
The whole CFLAGS settings should not be used, as far as I can see. I would 
*never* use -dead_strip either as long as I'm debugging compilation problems; 
that might strip code that is later used by plug-ins.


2) I have to say that I don't understand the following statement completely:

> I don't mind that some configure scripts have their own idea how they 
> are to be used, but I do mind that the rte configure takes some FLAGs 
> from the outside environment and does not follow what I've set at 
> starting the whole phase.

  -- That *FLAGs are taken from the environment is a wanted
     behavior, as with all programs that use GNU's configure framework.
     I.e., this is a feature and not a bug; we will not be able to change
     that, since we are bound to configure by the reuse of other projects'
     code.

  -- I don't understand the "does not follow" part. From the logs that you
     posted, the options that you specified were used during compilation.
     Perhaps I did not see it: Where did you expect which options to be used,
     and they weren't?


3) But the real showstopper are those syntax errors in shell commands during 
configuration or compilation. E.g., stuff like

  >     checking the maximum length of command line arguments... expr:
> syntax error
>     ../libtool: line 5306: test: : integer expression expected
>     ../libtool: line 5306: test: : integer expression expected
>     using piecewise archive linking...
>     ../libtool: line 5341: test: : integer expression expected
>     ../libtool: line 5341: test: : integer expression expected

This means that some Unix commands in configure and in libtool that work on 
dozens of Unix platforms don't work on Mac OS X. (This problem got reported by 
Simon as well.) I have no idea at all what causes this, that needs Mac OS X 
and/or Fink expertise to solve.

Maybe one needs to check if the same problem is still there with CLISP 2.38 
(downloadable from http://clisp.cons.org/). The test command sequence is:

     ./configure --with-export-syscalls --with-module=regexp \
         --without-readline --without-unicode
     cd src
     ./makemake --with-dynamic-ffi --with-export-syscalls \
         --with-module=regexp --without-readline \
         --without-unicode >Makefile
     make config.lisp
     make
     make check || make check


Cheers,
	Joachim

-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Joachim Schrod				Email: jschrod at acm.org
Roedermark, Germany

------------------------- Info --------------------------
Mac-TeX Website: http://www.esm.psu.edu/mac-tex/
          & FAQ: http://latex.yauh.de/faq/
TeX FAQ: http://www.tex.ac.uk/faq
List Archive: http://tug.org/pipermail/macostex-archives/