MathGroup Archive 1999

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: New position: MathLink Developer

  • To: mathgroup at
  • Subject: [mg16916] Re: New position: MathLink Developer
  • From: tobiasoed at
  • Date: Thu, 8 Apr 1999 02:32:30 -0400
  • Organization: Deja News - The Leader in Internet Discussion
  • References: <7e9ndq$>
  • Sender: owner-wri-mathgroup at

Jens,	I do not want to start a fight with you here since I respect what you
do a lot: I am aware that half of the answers to the questions on this
newsgroup come from you. but

In article <7e9ndq$3in at>,
  Jens-Peer Kuska <v-jkuska at> wrote:
> Hi Tobias,
> it is not a good idea to use *any* software without reading
> the manuals. You have probably never read more than the first 21 pages
> of "The Mathematica Book".

sitting right next to me. The only part I never read is the one about front
ends. I don't like them because the formatting of the expression doesn't
matter that much and chews up too much time. So I use the terminal like
interface. Unfortunately there is no history editing (like in bash or tcsh).
So I used jmath for a while, a frontend written by Jim Radford that does this
kind of stuff. In order to adapt it too my taste (and trying to fix bugs
related to interrupts - these are documented *nowhere*) I modified the code
for a while before rewriting the whole thing: the bugs are still there.

> Page 630--674 and 1256--1266 (Reference) of the Mathematica
> Book, Third Edition by Stephen Wolfram list almost all what you need to write
> a MathLink program.

as long as you do not try too fancy things: writing a function that you later
call from within a mathematica session using Install[] is Ok. But writing a
frontend is much less trivial. especially the interrupts (again...)

> There are additional documents on MathSource

printed out the tuttorial by Todd Gayley some time ago

> and many many programs. At least you can learn from my source codes at

guess what: it's in my bookmarks since quiet some time.

> how to use MathLink. Even for non-standard cases like loopback links,
> callbacks to the Kernel,  own event handling and parallel computing.
> Several "own" FrontEnds, like Leibniz
> have been written and so it seems to be possible, I had my own for
> version 2.x
> (on MathSource) and I will return to a 3. version of it soon.

I found one by S.C.Sheridan on mathsource. In the first few lines:

 * There is no way yet to send an interrupt to the kernel using
 * this frontend.

> As far as I remember the most questions about MathLink are answered in the
> news group.
> But to give a defined answer on has to know
> - the operating system


> - the compiler

HP-UX C compiler

> - the Mathematica version

Mathematica 3.0 for HP-UX PA-RISC

> only a few people using more than one operating system. The relative high
> costs of a commercial compiler make it nearly impossible to use more
> than one or two. So the probability that you find a person that can help you
> with MathLink on operating system X with compiler Y and Mathematica 2.2+Z
> is rather small. As long as one has not exactly the same configuration
> nobody can give you a defined answer.
> In the most cases a *defined* answer to a MathLink question can be only given
> when I see
>  a) the source code
>  b) can try a debugging session
> You can't not realy expect that anyone start with debugging your C-programs!

I don't ask for that - just a good doc would do.

> Not at Wolfram Research and not in the Mathematica news-group.
> You should take a look to other programing news-groups. The answers to
> questions like "I have a bug, I don't know why, here is a pice of C-code ..."
> are always vague. Every one that gives a "defined" answer to this kind
> of question has never used a compiler.
> Regards
>   Jens
> tobiasoed at wrote:
[ snip my original ramblings ]

-----------== Posted via Deja News, The Discussion Network ==----------       Search, Read, Discuss, or Start Your Own    

  • Prev by Date: Re: How to interrupt a running evaluation in MathLink
  • Next by Date: Re: ContourPlot: non-rectangular domains?
  • Previous by thread: Re: New position: MathLink Developer
  • Next by thread: Re: New position: MathLink Developer