Re: Not a timetable

[Dave Love <d.love@dl.ac.uk>]

>>>>> "Rat" == Stainless Steel Rat <ratinox@peorth.gweep.net> writes:

 Rat> The Emacs 20.1 and 20.1 binaries are ~50% larger than the Emacs 19.34
 Rat> binary (Linux/GNU system, GCC 2.7.2).  

I was thinking of a more relevant footprint like memory.

 Rat> The memory footprint is comesurately larger.

As in 4.8 MB for 19.34 v. 5.9 MB for 20.3 or 5.4 MB for 20.2 MBSK
freshly started under X on Linux as I see?  For a realistic sized
process, the fractional difference between 19 and 20 is considerably
reduced (not obviously important to me).  Gnus is one of the major
contributors to memory footprint in my Emacs, of course.

 Rat> Most of that code increase is MULE.

Probably so FWIW, but unmeasured.

However, arguing about the footprint misses the point of MULE problems
(such as are relevant to Gnus).

 KG> I think removing all features you don't use from all programs you
 KG> use is going to be a lot of work.

 Rat> Only becaue RMS decided to break his own "rules" and
 Rat> incorporated MULE in a fashion that prevents it from being
 Rat> removed.

What "rules", exactly?

 Rat> He could have done it right, witness XEmacs which has MULE as a
 Rat> compile time feature.  He decided not to do it that way.

While I don't like it, I can see reasonable arguments for it being
like that.  Note also that rms has reacted to complaints about the
MULE incarnation in 20.2.  20.3 deserves use.