The future of MathMap
schani
I apologize for not having been very responsive the past few months
and for not having pushed MathMap forward. The truth is, I have been
busy with lots of other stuff and have grown a bit weary of MathMap.
Several factors contribute to this:
* The difficulty and time intensity of building releases. Doing the
major Linux distributions is hard enough as it is, though the burden
is lessened a lot by the OpenSUSE build system, but Windows and MacOS
are huge time wasters and the work is not very interesting.
* Having to do almost all of it myself. I am sure I am mostly to
blame for this myself, not being very proactive in seeking
contributors.
* Even then, people still struggle just getting it to run, suggesting
that I'm very bad at packaging, documenting and/or not putting enough
time into it.
In short, as it is, I am currently, sadly not motivated to work on
MathMap.
There is a way forward, however! It's interesting, exciting, benefits
everybody, but I would need some help.
The gist of it is that I want to make MathMap into a web application
and I need somebody with a bit of web development expertise to help
build it. The advantages would be clear:
* Installing MathMap would be as simple as installing the proper web
browser (that is, Chrome - more on that later).
* Putting out new MathMap releases would be much less painful and
everybody would get them immediately, without any further
installation.
* Functionality like sharing MathMap filters could be built much
easier, by contributors who would not need to be fluent in C.
On the technical front, we'd use Google's Native Client (NaCl), which
allows native code supplied by the server to run on the client. The
way it would work is that the user writes some MathMap code (or,
later, uses MathMap Composer to write the code for them). That code
is then sent to the server which compiles it into code that can be
loaded by NaCl on the client, which is then loaded and presented to
the user. It would run at the same speed MathMap runs now and could
also take advantage of multiple cores.
What I need help with is the client side. Basic client web
development experience should be sufficient for a first usable rough
cut.
Any takers?
Mark
Owen
Wish you well.
Owen
Tom Rathborne
I am interested in the future of MathMap, but don't have a lot of
bandwidth available. A common story!
On Fri, May 13, 2011 at 04:18:21AM -0700, schani wrote:
> I apologize for not having been very responsive the past few months
> and for not having pushed MathMap forward. The truth is, I have
> been busy with lots of other stuff and have grown a bit weary of
> MathMap. Several factors contribute to this:
>
> * The difficulty and time intensity of building releases. Doing the
> major Linux distributions is hard enough as it is, though the burden
> is lessened a lot by the OpenSUSE build system, but Windows and
> MacOS are huge time wasters and the work is not very interesting.
>
> * Having to do almost all of it myself. I am sure I am mostly to
> blame for this myself, not being very proactive in seeking
> contributors.
>
> * Even then, people still struggle just getting it to run,
> suggesting that I'm very bad at packaging, documenting and/or not
> putting enough time into it.
Yes, it really needs to work with GNU autoconf and not the current
tarball full of hacks and assumptions. Also, whenever I have dug into
the code, I have concluded that only you can understand it well enough
to update it.
As you know, I do weird things with MathMap, so my interests lie in a
distributed version of MathMap that I can run on (for example) a
Hadoop cluster. My current solution is a shell script which lets me
run 1 instance of MathMap per animation frame. I am satisfied with
that.
I think MathMap needs to be split into "libmathmap" which encapsulates
the various ways that expressions are executed (interpreter, gcc,
LLVM, etc.) and then various frontend clients (CLI, GIMP, MacOSX,
Chrome).
I like the idea of making MathMap more accessible, though I don't need
it. I am wondering what use cases you see for the Chrome client.
Would it just be for playing around? Would people embed MathMap
animations in their web pages? Could I apply MathMap expressions to an
image and upload the result directly as a JPEG.
I guess an auto-Droste-Effect webpage would be pretty cool. :)
Cheers,
Tom
--
-- Tom Rathborne ------------------------------------- Ad-hoc Research Robot --
When they tell you to grow up, they mean stop growing.
-- Tom Robbins
