Re: ITP ptolemy?
On Fri, Jun 15, 2001 at 10:12:39PM +1000, Hamish Moffatt wrote:
> I'm thinking about packaging ptolemy. This is a decent-sized
> application useful for simulating electronics system, particularly
> DSP. It's pretty cool stuff. The license is BSD.
>
> There's a complication though.. users can develop their own bits of
> code (called stars, which become part of a galaxy), and their code is
> dynamically linked to the ptolemy binary itself at runtime. So the
> upstream authors say the user's code needs to be compiled with the
> same version of gcc as ptolemy itself.
isn't this what shared librariers and the concept of APIs are for?
seems to me that the real long-term solution is for the ptolemy authors
to design and implement an extension API, a plug-in architecture.
of course, that would be a lot of hard work.
> There could also be a problem with it requiring a customised
> version of tcl.)
yuk! this kind of forking is fairly common with tcl stuff, which is the
main reason i lost all interest in tcl & tk a few years ago.
> Suggestions? A ptolemy-gcc package which installs itself into
> /usr/lib/ptolemy? That's the preferred option by the upstream
> authors; their precompiled binary packages include gcc and tcl
> binaries.
at a guess, i'd say ptolemy would have to be at least 8-10 years old
then. this kind of bundling was very common in the days before free
*nixes and the net made it a reasonable assumption that such tools would
be available and would/should be updated separately.
it's always a severe PITA to extract this kind of program from its
bundled dependancies.
> Input welcome. I haven't sent in a wnpp bug for this yet
> as I might change my mind!
how about a wrapper script for executing ptolemy and/or "stars"? it
would compare the version of gcc used to compile gcc with the version
of gcc used to compile the stars...recompiling the stars whenever
necessary.
simultaneously, encourage and assist the authors to begin a
modernisation program so that it meets modern expectations of how a
program should work and interoperate with other programs, how it should
be extensible, and how it should be distributed.
craig
ps: it's easy to be a backseat driver...i'm not planning to do any of
this work so feel free to dismiss my comments as impractical idealism.
--
craig sanders <cas@taz.net.au>
Fabricati Diem, PVNC.
-- motto of the Ankh-Morpork City Watch
Reply to: