Re: BLT Graph Derivative



Hi Bob,

good to see things starting now!

We're just about to start requirements definition. I'd appreciate any
input or feedback that the Tcl community would like to provide.
....
I know that a number of folks have been talking or asking about this
for some time, and I believe that Torsten Berg has actually done some
coding.

Yes, I have started ripping the vector command out of BLT and making
it work independently under Tcl 8.4 and 8.5. Regard this as a case
study to see how complicated it is and to learn how BLT is coded and
organized internally. My approach and the resulting source code may
not be perfect and some things may even be "un-C-ish". But I realized
the whole work could be done in comparatively few days.

You can find some more information about the degree of success and the
implementation in the "BLT overhaul" thread here on c.l.t. I have put
the current source code here:

http://tcl.typoscriptics.de/misc/bltVector.zip

Some of the things I have changed starting from the original BLT
sources are:

1. removed the use of the bltAlloc package to now use ckalloc and
ckfree.
2. removed the use of the bltHash package to now use Tcl_Hash... (this
might result in some problems for 64bit systems, I am not sure.
Integrating the TIP 69 into the core would then overcome this
potential problem).
3. changed the string based Tcl commands to Tcl_Obj commands where
applicable.
4. avoid the use of the internal Tcl C API to make vector stubs-
enabled.

I can (and will) give some more specific information when I'm home
again next week (I'm attending a symposium right now and there's not
much time here).

Torsten
.