RE: apr-iconv? [OpenSSL? zlib? ldap?]

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

At 02:25 PM 8/1/2002, you wrote:
>On Thu, 1 Aug 2002, Sander Striker wrote:
>
> > > Is "apr-iconv" required now when building from CVS on Linux?
> >
> > No, only for win32 builds.
>
>For the record, I think it's a terrible idea to just sort of sneak this in
>there (in as much as it came to a huge surprise to someone who hasn't been
>able to keep up with email for the last week-and-a-half or so).  I guess I
>missed the discussion on why it was necessary... I thought we were just
>improving *detection* of it.  ??

Agreed to a point, disagree to a point.

If the user must rebuild/use an alternate binary to support iconv... it means
that the usual distro of httpd will vary significantly from the versions of apr
required by many i18n applications, such as subversion.  That stinks.

If we were able to plug in dynamically, based on the presence [or absence]
of libiconv, or libapriconv, I would be much, much happier.

OTOH, building without any iconv libraries /w/c/ould also be a good thing.
In win32's case, that could be a disable, rather than an enable toggle.

But any which way you cut it, we need some configuration for win32.
That includes the presence and location of openssl libraries, zlib sources
and the iconv libraries.

Expect me to devote my Sunday afternoon to developing just such an
awk or perl script, that would go out and update the .dsp and apache.dsw
files for the correct libraries and modules that may be built based on these
libraries.  This means we won't need the oddball "invoke the .dsp from the
makefile.win from the apache.dsw workspace".

For now it will just determine what's going on in httpd-2.0/srclib/ ... later
we can add additional flexibility with paths.

Bill