Ned Konz <ned at bike-nomad.com> wrote:

> I just went to try out Filtering Celeste, which requires Large Lists, 
[snip]
> I wonder whether these methods could be packaged separately and added 
> to the 3.6Full load script for those people who want to load the 
> older packages.
>
At some point we have to accept a break in supporting every little bit
of past cruft. If we clean up the system we will inevitably break
packages or fileins that relied on the old code. The really
sensible answer, obviously, is to revise the packages to work with the
new world which equally obviously involves work. Somewhere we need to
work out where to draw the line and how to make tools to help us do
things better.

I guess a really nice property of SM is that we can find all the
packages it refers to, so they can be tried and fixed. All other
packages can be declared 'unknown' and they are purely the
responsibility of authors and any hopeful users. This improves the value
of being listed on SM.

Every now and then one does have to 'burn the disc packs' in order
to advance; the new image format would render lots of image segments and
projects useless for example. The current prototype of declarative pools
would cause big problems with any code that had used the current style
of filing out pool dictionaries.

I certainly don't claim to know where to draw the lines right now but I
do know we need to start deciding sometime soon.

tim
--
Tim Rowledge, tim at sumeru.stanford.edu, http://sumeru.stanford.edu/tim
Useful random insult:- Been napping in front of the ion shield again.