According to Geoff Hutchison:
> On Wed, 11 Aug 1999, Gilles Detillieux wrote:
> > It's an ugly hack, but it works. BSDI users also need to resort to a
> > hack to use the C library regex code instead of htlib/regex.{h,c}.
> > We need a better solution.
>
> I don't know of a good solution for the header files. For actual *code*,
> autoconf macros can be used to add particular files to a target list to
> ensure it's compiled. This is much like the autoconf tests that add to
> LIBS.
Loic's solution for the strptime.c/langinfo.h problem will probably do
the trick in most/all forseeable cases. If not, it's tweakable, as I
suggested in my previous message.
The BSDI regex problem is a bit trickier. It's not enough to test for
the presence of regex.h in /usr/include, and regcomp in the C library.
As I said before, on my Red Hat 4.2 system, it has these, but when I
build htfuzzy using these, rather that the source in htlib, it crashes
quite badly - so much so that I can't even get an intelligible stack
backtrace to find out why it's dying. So, on BSDI, the fix is to use
the system's regex code, but for other systems, the bundled regex code
seems to be called for. I don't know how you'd get autoconf to test
for this.
There may still be other, more subtle problems with the new regex code.
One user complained that the endings fuzzy matching stopped working on
his system (Solaris, I think), when he switched from 3.1.1 to 3.1.2.
Hopefully we can get some more thorough testing of this in before
releasing 3.2.0, because we're relying on regex more than ever now.
> Does anyone have a good solution for including either system headers or
> headers in a particular directory?
>
> (Come to think of it, that may be it--we might put the system replacement
> headers in a particular directory and include that as a -I as necessary.)
Wouldn't that require a separate directory for every header file we
want to selectively include? I think that what we should do instead is
make sure our own headers have unique names, that won't conflict with
system headers. Then we can put tests like this in the C/C++ source
that includes these headers:
#ifdef HAVE_FOO_H
# include <foo.h>
#else
# include "htfoo.h"
#endif
--
Gilles R. Detillieux E-mail: <[EMAIL PROTECTED]>
Spinal Cord Research Centre WWW: http://www.scrc.umanitoba.ca/~grdetil
Dept. Physiology, U. of Manitoba Phone: (204)789-3766
Winnipeg, MB R3E 3J7 (Canada) Fax: (204)789-3930
------------------------------------
To unsubscribe from the htdig3-dev mailing list, send a message to
[EMAIL PROTECTED] containing the single word "unsubscribe" in
the SUBJECT of the message.