Re: New Emacs with GTK!

[Alan Shutko]

Niels Freimann <nfreimann@firemail.de> writes:

> To separate a program into different layers is a paradigm of modern software 
> development. You must know that.

Have you looked at the redisplay code?  There's not much code there
which is specific to a given windowing system.  Separating Emacs into
a libemacs, TTYEmacs, OSXEmacs, GTKEmacs, W32Emacs would result in a
really big libemacs, and really, really small window-specific
versions.  That would provide a negligible size advantage downloading
the source, an even smaller advantage in the binary, and basically no
advantage in memory use.  (Probably even a decrease in performance on
some platforms like x86 Gnu/Linux, since a shared libemacs would have
lost a register to -fPIC.)

The cost for these minimal benefits would be more work for the
development team, since APIs would have to be made more rigid since
someone changing an API couldn't just update all the calls in the
various frontends.  Or it would mean users would lose out, because
the various frontends would no longer be truly compatible.

Objectively, what is the case to be made to do this?

-- 
Alan Shutko <ats@acm.org> - I am the rocks.
Looking for a developer in St. Louis? http://web.springies.com/~ats/
Don't underestimate the bandwidth of a station wagon full of mag tapes.