David Burrows wrote:

> Oh, so you like bloated binaries which waste vast quantities of ram and
> hard disk space?  I'd like to hear a good reason why.  Shared libraries
> are an infinitely useful feature.  And lumping in fat static libs to every
> executable on your hard disk is rather inefficient, regardless of whatever
> argument you might have against shared libs.

There is a good reason for static libraries and I am saddle sore from having
to deal with that particular issue.  When you are doing software for an operating

system that manages to change a lot, you have to be concerned with versions
of shared libraries.  A project I am involved with has a port for Linux.  I asked

the kernel version and niavely sent out a copy that was built using shared
libraries.  I got complaints about core dumps that I could not reproduce.  When
I sent a statically linked version, the problems went away.

I would agree with David and be in favor of shared libraries, but only on a
relatively stable platform that is not constantly evolving.  I am speaking about
a difference from like a Solaris -vs- somthing build with GCC something-or-other
which has the same impact in the NetBSD community.  The one thing that NetBSD
has going for it, is that there is only one distribution, which simplifies things
radically.
Here I only worry about testing things such as 1.2C, G or 1.3.

I guess I would summarize myself in saying that things tied to an O/S would
have advantages to using shared libraries, but external things would be better
statically linked so the regression testing needed is kept at a minimum.

Andy

> > And please stop calling Berkeley UNIX(R) "vaporware". It exists, and has
> > existed in a reasonable shape since 1986 when 4.3BSD, the first version with
> > satisfactory ARPA Internet support, was released.
>
> Can I boot this on my VaxStation 3100? :)
>
> >
> > Sincerely,
> > Michael Sokolov
> > Phone: 440-449-0299
> > ARPA Internet SMTP mail: sokolov@alpha.ces.cwru.edu
> >
>
> Regards,
>
> Dave.