Re: tlug: help! intallation of Wnn4

["Stephen J. Turnbull" <turnbull@example.com>]

>>>>> "Scott" == Scott Stone <sstone@example.com> writes:

    Scott> On Sat, 16 May 1998, Takeshi Masuda wrote:

    >> Please, anyone who has successfully installed the RPM package
    >> of Wnn4 help me.

Wnn is a bear to work with (or worse, a comparison with Boris Yeltsin
comes immediately to mind, probably favoring Boris's kind and amiable
personality).  Canna is much more user-friendly.

    >> site, and made an user "wnn", and then installed it. Next, I
    >> got the source of xemacs and tried to compile it with mule and

What version of XEmacs?  Currently supported versions include 19.15,
19.16 (but they don't support Mule), 20.3 and 20.4.  21.0 is in beta,
and there is some noise about a 20.5 (the 21.0 package system will
make transition annoying for some people).

    >> Wnn support, but I couldn't. Jeff Miller, who read my article
    >> at comp.emacs.xemacs, told me that I couldn't compile xemacs
    >> because I didn't install Wnn4 correctly.

This is probably true.[1]  As already discussed, you need both the Wnn
development libraries and the Wnn header files.  XEmacs sometimes gets
a little confused about the location of X-related libraries,
especially ones like Wnn that don't bother to try to be compatible;
you may be able get those libs found by using
`--site-prefixes=/usr/local' on the configure command line.  If that
doesn't work, you may need to use `--site-includes' and
`--site-libraries'.  Do not use --site-runtime-libraries; you will be
sorry if you try that.

    >> in /etc/ld.so.conf, and typed "ldconfig", but I still can't
    >> compile xemacs.

This doesn't help much; you need to tell XEmacs configure, and thereby
ld, where the libraries are; ldconfig tells ld.so where they are.

    Scott> Did the wnn4 rpm put anything into /usr/local/include?  You
    Scott> might need to make those available to the compiler for
    Scott> xemacs as well.  I've also heard that xemacs doesn't build
    Scott> right on glibc systems at all.

This is a lie.  Every XEmacs beta and release since 19.15 and 20.1, 18
months at least, has built on several glibc systems.  Probably about
half of the beta testers, maybe more, who build on Linux build with
glibc (and egcs, but that's another story) as of 21.0.  In particular
the head maintainer builds with Mule, Wnn, Canna, and SJ3 for every
beta on a glibc system.

It is true that it is almost impossible to build anything non-trivial
on many "stock" glibc systems.  But this is the fault[2] of the
incompatibility of libc5 with libc6, and the incompatibility of
glibc with everything, not at all a problem with the application
software.

I haven't tried very hard, but I have yet to see a glibc
incompatibility that did not manifest itself with both XEmacs and the
FSF's release of Emacs---of course no incompatibilities with Wnn will
manifest with current FSF-designed Emacsen, since the FSF has
abandoned support for all backend servers.[3]

HTH

Footnotes: 
[1]  It's not just you.  Gotta love RedHat.

[2]  Strictly speaking, the fault is of the distributors who release
beta systems and charge full price for them, not of the libraries
whose developers are busting butt to make them work, and work right
for the right reasons for a change.  You will note that Debian still
has not released a glibc-based system as its default configuration---
they evidently think "free" is too much to charge.

[3]  Well, "fgrep -i wnn *" gives some hits on the word "knownness"
[sic], but none on Wnn, in emacs-20.2 and emacs-20.2/src.



--------------------------------------------------------------
Next Nomikai: 15 May Fri, 19:30 Tengu TokyoEkiMae 03-3275-3691
Next TLUG Meeting: 13 June Sat, Tokyo Station Yaesu gate 12:30
Featuring Stone and Turnbull on .rpm and .deb packages
--------------------------------------------------------------
Sponsor: PHT, makers of TurboLinux http://www.pht.co.jp