> Lest ye think only Asger and Alejandro have opinions on this
> subject...
Ahh, no. I think everybody have an oppinion about this. I'm just wasting
more time with talk than with code.
> 1) First, let's get one thing clear: comment-embedded LyX commands
> won't fly.
Agreed.
> 2) Second, let's get another thing clear: parsing a fixed subset of
> LaTeX is not hard.
If the subset is chosen carefully, this is trivially true ;-)
For an interesting subset, this claim might not be true.
> 3) LyX3 --- whatever it is --- should not be editable casually.
It could be, but it's not supposed to be. The key thing is to let the
people that do it know that they are messing with fragile stuff.
> 4) We should choose that combination of LaTeX and DocBook which is
> most powerful for the features we need.
>
> I must admit, I hate the current LyX file format.
>
> Now, what I mean by combination is a "hybrid" format.
Have a look at the stuff that is produced by the development version. It
is getting much better. If you look carefully, the format is not that
bad. It has a few key-elements that are wrong, but *when* these are
fixed, the format is pretty good (and robust).
I think we should pursue this direction instead of dumping all the code
that we have. Fix the existing code instead of creating new code, and
then fix that.
> 5) The two criteria we need to balance are Extensibility and
> Parsability.
The current format is extensible and parsable: We have proof in the
existing version. Also, these two proporties are not working against each
others. It is perfectly obtainable to get something that is easy to parse
and extendible at the same time.
XML is one primary example (ok, it's not easy to parse, but that's not
something conceptual. That's just because it's big.)
The LyX format as defined by the development version in a few months (;-)
is another example.
Greets,
Asger
