> A nice solution would be a scripting language which is small enough
> to be included in the distribution without major bloat. Does such a
> beast exist?
The best candiate for this solution is SIOD: Scheme In One Defun.
It's small and lean, and easily customized for what we need.
The main problem is the issue of the syntax of the language,
which will effectively limit the target user group to the
current developers and maybe twenty more people.
The main benefit for us would be to help ourselves as developers,
because we could implement parts in Scheme.
However, I'd prefer to use a simpler language with easy syntax
so that the target market is the user that knows how to use
LyX (i.e. not the complete novice, but the big middle group).
This problem of syntax is not easily "customized" away in
Scheme.
The other alternative is to design a mini-language for ourselves.
While this would be a fun project (I have designed and implemented
several toy languages so far), this would take away too many
resources for other things, and this might be a bad investment.
But unless we find something good, this might be what we'd have
to do. If we do that, I anticipate that we will have lots and
lots of discussions of how the language should be designed,
and maybe too much disagreement. Everybody have their own
ideas of a perfect language, and this will not synch easily.
Therefor for this approach to be succesful, I think it would be
more realistic if we depart a bit from the normal "democracy"
development and give a single person a leader position with
the final say. To make it more fair, we could pass this
leader role around in a round-robin fashion, but I feel
that it is important that somebody will have the power
to settle the discussion without the others to be too
angry and disappointed.
If we could agree on such an arrangement, it might not be
entirely out of the question to succeed in this project,
if it is done in an incremental fashion: First, we design
and implement the absolute basic features of the language.
Later, we extend the language with more advanced features,
data structures and libraries.
But I still think that it would be best if we could find
an existing language that fulfills all our requirements.
> If we use the second solution, I guess we should offer support for any
> script language. Of course, we need an official one if large parts of
> LyX use it.
And this would be with the caveat that other scripting languages
than the official one would be potentially insecure.
Greets,
Asger
