Re: [Texmacs-dev] Mathematica Plugin

[Joris van der Hoeven]

Thanks Max; that seems the right way to do things; if someone has time...
In the links there is more info about building mathlinked executables.


On Wed, Apr 25, 2012 at 02:04:20PM +0200, Gubinelli Massimiliano wrote:
> It is possible to create an executable inside mathematica
> 
> http://reference.wolfram.com/mathematica/CCompilerDriver/ref/CreateExecutable.html
> 
> this maybe helps with the process of building the plugin code. Instead of 
> using a script, directly compile it inside mathematica.
> 
> my 2c.
> 
> max
> 
> 
> 
> On 25 avr. 2012, at 13:08, Joris van der Hoeven wrote:
> 
> > Dear Philippe,
> > 
> > On Wed, Apr 25, 2012 at 09:32:05AM +0000, Philippe Joyez wrote:
> >> Sorry to disappoint you but I'm afraid that getting a working mathematica 
> >> plugin
> >> "out of the box" for new users is not going to happen before the summer. 
> >> However
> >> a user with some motivation and knowledge can make it to work. Let me 
> >> explain
> >> why.
> >> 
> >> What's broken is not Andrey's mathematica plugin code itself, it's the 
> >> one-time
> >> script that runs the first time the user wants to connect to mathematica. 
> >> That
> >> script *compiles* the plugin code, linking it to the actual mathlink 
> >> library on
> >> the user's system and puts the resulting executable in the user's Texmacs
> >> directory. Since the time when Andrey wrote the script, there has been 3 
> >> major
> >> releases of Mathematica and the added support of MacOS and Windows in 
> >> Texmacs,
> >> that all broke that installation script in various ways.
> >> 
> >> Now, fixing the script so that texmacs would in all cases transparently 
> >> connect
> >> to mathematica without any user intervention seems unlikely to happen for 
> >> the
> >> simple reason that a compiler may not even be available on the user's 
> >> machine.
> >> So manual installation of the (OS-dependent) toolchain would be necessary
> >> anyway. Furthermore the new script(s) would have to handle many OSes and 
> >> many
> >> possible mathematica versions, which means someone should have access to 
> >> all
> >> (OS-mathematica version) configurations... but who would that be?
> >> 
> >> Presently I see no solution other than providing good how-tos on the
> >> compilation, or at best platform-specific scripts. What I posted last year 
> >> in
> >> the users' list 
> >> (http://comments.gmane.org/gmane.editors.texmacs.user/7409) is
> >> how to perform this one-time compilation in linux, which isn't really a big
> >> deal. In Albufeira, Miguel worked out the way to do this on MacOS for 
> >> Bertrand.
> >> If I understood correctly, it was not as easy as in linux and that should
> >> probably be written into a howto as well. Then it would certainly be nice 
> >> to
> >> centralize these instructions at some place (the famous wiki? a forum with 
> >> a
> >> dedicated "plugins" section?) so that they could be maintained up-to date 
> >> with
> >> user contributions for new setups, new versions, etc. 
> > 
> > What I propose is that those who are willing to work on the howto's
> > also try to fix the installation script at least in those cases when
> > they can make it work.
> > 
> > I agree that this is not nice to program, since it might involve tests
> > on both the OS and the version of Mathematica. However, I think that
> > this is really how a novel user would expect things to behave.
> > 
> > The real pain here is that Mathematica and Maple provide so called
> > 'Open' interfaces which we are not allowed to link in to our software.
> > So there is not much else that we can do than compile things on the fly.
> > 
> > Best wishes, --Joris
> > 
> > 
> > _______________________________________________
> > Texmacs-dev mailing list
> > address@hidden
> > https://lists.gnu.org/mailman/listinfo/texmacs-dev
> 
> 
> _______________________________________________
> Texmacs-dev mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/texmacs-dev