Re: On Compatibility

[Jonathan S. Shapiro]

On Tue, 2005-10-25 at 21:05 +0200, Alfred M. Szmidt wrote:
>    It is hypothetically possible to get all of the configured files
>    right by hand, but in practice we didn't have enough years in our
>    expected lives to do it. Perhaps you can show us how to do
>    it. Assistance would certainly be welcome.
> 
> Hack configure.ac, generate configure on another platform, continue
> hacking untill port is done.  No need whatsoever to have autoconf on
> the target system (you could even cross compile!).
> 
>    Sure it is. Consider bash.
> 
> So lets consider bash, what is wrong with it?

As I have already said, "legacy" does not mean "bad", and there is
nothing wrong with bash. But in the large scheme of things, bash simply
does not have a large enough user base to be of primary importance.

> Emacs must be complete and utter crap if you base things based on how
> long they have existed... And you'd be suprised how similar todays
> Emacs is to the Emacs that ran on ITS.

I probably wouldn't be surprised at all. Once again, I did not say that
legacy is bad. Like bash, the user base for emacs is tiny.

One does not design successful systems by focusing on the rather strange
desires of a small number of developers. One asks: how do I best support
the end users, and within that constraint, how to I keep the developers
interested? End users do not use emacs!

>    Or m4. Hell, even Stu Feldman thought m4 was crap, and he wrote it!
> 
> Care to give an actual reason why m4 is crap?

Nope. I'll leave that to Stu. But M4 is *both* legacy *and* crap.

shap