[tug-summer-of-code] GSoC participation

[Christoph Hafemeister]

Hello,

as I am also interested in participating in this GSoC project, I have
been following the discussions here.
I agree with Jonathan that user experience should be the main focus.
The existing scripts should be tweaked and adapted to work flawlessly
with IE7, 6 & 5, Gecko based and KHTML based browsers.

For client side error handling of Tex code I like the idea of using a
dictionary of Tex commands MathTran understands for finding unknown
commands and catching "! Undefined control sequence." errors. DHTML
could be used to present alternatives whenever an unknown command was
detected. Other errors could also be detected by JavaScript but it
should be avoided to force the user to change existing code (like
adding brackets around single element superscripts).

When using AJAX and the error log of the server side script, more
effort could be put into interpreting the error messages and trying to
present them to the user in a helpful way (syntax highlighting,
original error message). This would also keep the JavaScript source
smaller.

>  The HTTP headers set by MathTran set a time to expiry of, as I recall,
>  10 minutes.  This causes the BROWSER to cache the image.  I'm told that
>  for this purpose it would be better if the MathTran interface did not
>  use a query string (GET request).

I believe GET is fine, when using HTTP/1.1 or higher with the right
headers. Responses to the POST method might not be kept by most
caches.


How far is the work on the MathML-to-TeX translator? When this
translator is working, the JavaScript code could be extended to send
MathML encoded formulas to the server and replace them with the
corresponding pictures. The decision whether to convert MathML to Tex
generated images could be done by the script via browser and/or plugin
detection or by the user via bookmarklets.

Best regards,
Christoph