[ clisp-Bugs-1489936 ] regex.h gets clobbered by Glibc's regex.h

[SourceForge.net <no...@so...>]

Bugs item #1489936, was opened at 2006-05-16 22:21
Message generated for change (Settings changed) made by sds
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=101355&aid=1489936&group_id=1355

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: regexp
Group: build problems
>Status: Closed
Resolution: Works For Me
Priority: 5
Submitted By: Eitan Levi (drskrud)
>Assigned to: Nobody/Anonymous (nobody)
Summary: regex.h gets clobbered by Glibc's regex.h

Initial Comment:
I had some problems compiling Clisp 2.38 since the
build would constantly fail while trying to compile the
regexp module, complaining that a number of macros were
undefined identifiers. Some careful investigating
revealed that gcc was #including the system regex.h
file (from /usr/include) which is part of Glibc (I'm
using version 2.3.6). The Glibc version of regex.h does
not define such things as REG_TRANSLATE_TYPE, which
caused the build to explode.

Going to clisp's regexp/regex.c file and changing the
line that said #include <regex.h> to #include "regex.h"
solved the problem by having the build rely on the
local copy of the header file. Even though there *was*
a -I. option, I think GCC (I'm using version 4.0.2)
still gave /usr/include higher precedence.

----------------------------------------------------------------------

Comment By: Eitan Levi (drskrud)
Date: 2006-07-21 01:16

Message:
Logged In: YES 
user_id=871542

This bug report is now marked as "pending"/"works for me".
This means that we think that we cannot reproduce the problem
and cannot do anything about it.
Unless you - the reporter - act within 2 weeks
(e.g., by submitting a self-contained test case),
the bug will be permanently closed.
Sorry about the inconvenience - 
we hope your silence means that 
you are no longer observing the problem either.


----------------------------------------------------------------------

Comment By: Eitan Levi (drskrud)
Date: 2006-07-21 01:16

Message:
Logged In: YES 
user_id=871542

I switch distros as often as most people switch desktop
wallpapers. I like experimenting -- anyway it turns out
GoboLinux was full of these kinds of little problems. The
regex.h was indeed getting clobbered but I suspect the
culprit was some weird GoboLinux configuration issue w.r.t. GCC.

I've since tried out ArchLinux and now I'm on Ubuntu Dapper.
CLisp works great on both ;)

----------------------------------------------------------------------

Comment By: Sam Steingold (sds)
Date: 2006-07-20 16:40

Message:
Logged In: YES 
user_id=5735

This bug report is now marked as "pending"/"works for me".
This means that we think that we cannot reproduce the problem
and cannot do anything about it.
Unless you - the reporter - act within 2 weeks
(e.g., by submitting a self-contained test case),
the bug will be permanently closed.
Sorry about the inconvenience - 
we hope your silence means that 
you are no longer observing the problem either.


----------------------------------------------------------------------

Comment By: Sam Steingold (sds)
Date: 2006-07-20 16:39

Message:
Logged In: YES 
user_id=5735

at any rate, the clisp regexp files come from gnulib verbatim
so this bug - if indeed it is one and not a misconfiguration -
should be reported to gnulib.

----------------------------------------------------------------------

Comment By: Eitan Levi (drskrud)
Date: 2006-05-20 04:08

Message:
Logged In: YES 
user_id=871542

If that's the case, then it's likely a problem particular to
GoboLinux... It's most likely a GCC configuartion error, as
you mentioned, given that GoboLinux has an "unusual"
directory hierarchy.

Either way, I'm sorry I can't confirm what was going on
since I've since switched to ArchLinux, on which Clisp
builds fine.

----------------------------------------------------------------------

Comment By: Jörg Höhle (hoehle)
Date: 2006-05-19 10:41

Message:
Logged In: YES 
user_id=377168

Could you please investigate GCC's behaviour closer?
-I. should take precedence -- normally.
However, the manpage also says:
If a standard system include directory, or a directory
specified with -isystem, is also specified with -I, the -I
option will be ignored.
E.g. there's potential for misconfiguration of GCC here: -I.
could be viewed as a system dir and ignored.

Sadly, I haven't managed to find an option to have gcc/cpp
dump its default include directories. These did not help:
 gcc --target-help
 gcc -print-search-dirs


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=101355&aid=1489936&group_id=1355