On 22 Oct 2000, Lars Gullik Bjønnes wrote:
> "Asger K. Alstrup Nielsen" <[EMAIL PROTECTED]> writes:
>
> Do you know if anyone has mentioned xtl on the boost list? Perhaps
> that should be done?
>
> Of course to include the xtl in boost the lisence has to change.
Asger sent an email to the XTL list advocating XTL's submission to Boost.
It was the first email on the list for about 2 months or more. And I
still haven't seen a reply.
> (btw. I have serious dubts if xtl is the right thing for lyx
> internals.)
I'm considering doing some work on XTL to cut down it's compiler
requirements and try to do something about text import (so it supports XML
style text exports). That alone should stop 90% of complaints about XTL.
Cutting down compiler requirements will definitely affect its capabilities
however we should be able to get a nice subset that works on most
compilers since gcc-2.8.1.
BTW, XTL buys us lots of stuff for eXternal access. It is also convenient
for internals. The C/C++ Users Journal article I mentioned the other day
has several features I think could be used to reduce the compiler
requirements of XTL (rtti is still necessary though I think). The CUJ
article has what could be described as an "XTL for C++ to C++" in that the
data isn't actually externalised but wrapped up with type info and stored
in a fancy variant_t class -- basically a glorified void*.
XTL could be the backbone for providing CORBA access to our internals (the
data exchange being handled by XTL). Any half decent scripting language
should also have XDR or GIOP libraries written for it (like Perl and
Python do) so XTL can provide the scripting languages with access to
internal data in a format that looks native to them -- avoiding all that
silly custom string handling.
Allan. (ARRae)
