Will Robertson wrote:

> I think for now assume that anything that's written with the expl3
> package(s) is to be used in a LaTeX2e environment. I guess I consider it
> a bit like the etoolbox package: it provides package writers with better
> tools  to do their job. Worry about LaTeX3 when there's actual talk of
> formats being built that drop LaTeX2e support.

The difference (in the short term) is that etoolbox is "stable".  So I
can use that in production packages and not worry about the macros
available altering in the future.  Anything I write with expl3 is itself
experimental as a result.

[Of course, I note that etoolbox has had some renames going on.
However, it is "released" and most of the changes were pretty soon after
it came out.  So the general point stands.]

> (Err, agreed, everyone?)
> 
> This way you can minimise the code you have to write (since whichever
> LaTeX2e package provides it) and it will perhaps become clearer which
> (currently external) packages do need to be pulled into the kernel.

Here is the point, really.  I can write what I like using expl3, but I
can't release it as production.  So everything has to be done twice,
once in standard TeX/LaTeX and once in expl3 (if I want to do the later
at all).  That will change when the current l3 packages reach a point
where I can rely on them not changing, where I can then write new stuff
in l3 + support as needed from LaTeX2e packages, with the aim of
minimising the later.

(By the way, I can't see that any LaTeX3 kernel can possibly support 2e
packages.  Surely the entire thing will be structured so differently
that writing the necessary hooks will be too much work.)
-- 
Joseph Wright