Re: ncurses-5.1-20001014.patch.gz

[Thomas Dickey]

On Sat, Oct 14, 2000 at 11:42:21PM -0500, Albert Chin-A-Young wrote:
> On Sat, Oct 14, 2000 at 11:12:37PM -0400, Thomas Dickey wrote:
> > 20001014
> >     + test/repair some issues with libtool configuration.  Make
> >       --disable-echo force libtool --silent.  (Libtool does not work for
> >       OS/2 EMX, works partly for SCO - libtool is still very specific to
> >       gcc).
> 
> FYI, libtool 1.3.5 works fairly well with vendor compilers too (I have
> not tested the libtool support in ncurses yet). The only problem with
> the 1.3.5 version is C++ which does not work well with vendor

yes, that's what broke with SCO.  EMX just plain doesn't work.  I test
compiled on Slackware 3.6, Redhat 6.1, Debian 2.1, FreeBSD 3.1, Solaris
7, cygwin b20.1, OS/2 EMX, and SCO OpenServer.

> compilers. The multilang branch fixes this though. As ncurses uses
> both C/C++ (though C++ can be turned off), maybe the multilang version
> should be considered.

something like that.  Since it's a late change, my main aim was to ensure
that incorporating it would not break anything I'm doing currently, and
that it works on the platforms where people are most likely to use it
(no point in adding a feature only to get more bug reports).
 
> Also, saying this without having looked at the --with-libtool option,
> do the version numbers of the ncurses library match up if using
> --with-libtool and without?

I haven't explored it completely:  building with libtool takes a lot longer
than building the way I've been doing, perhaps 10 times, so I've done only
sample compiles on my local platforms).

They seem similar -- for Linux and FreeBSD (major.minor).  That of course
doesn't port well, so I'm curious to see how libtool does handle systems
where the minor version is irrelevant.

I can test-compile later on DU 4.0 and IRIX 6.2 (perhaps that HPUX 9.x box
I compiled on before is still running).
 
-- 
Thomas E. Dickey <address@hidden>
http://dickey.his.com
ftp://dickey.his.com