The Higher Education and Research forge

Home My Page Projects Code Snippets Project Openings Développement de WIMS
Summary Activity Forums Tracker Tasks Docs Surveys News SCM Files Listes Sympa

Forum: developers

Monitor Forum | Start New Thread Start New Thread
RE: mathml mathjax [ Reply ]
By: Bernadette Perrin-Riou on 2016-10-27 08:36
[forum:17711]
OK. Perhaps we will have to check from time to time if safari or another one improves their native mathml !

Bernadette

RE: mathml mathjax [ Reply ]
By: joke evers on 2016-10-26 16:15
[forum:17710]
And do we agree on the issue of browser-engine-detection ?
e.g. only real gecko-engines should use native mathml...the rest should use Mathjax

RE: mathml mathjax [ Reply ]
By: Olivier Bado on 2016-10-26 14:36
[forum:17709]
it seems ok in Safari too (it's not possible to go to the line).
But it is still a crutch, as it will not have exactly the same behavior as the input tag (where the enter key submits the form)

But it's still better than a completely non working issue, I must admit ;)

RE: mathml mathjax [ Reply ]
By: Bernadette Perrin-Riou on 2016-10-26 11:08
[forum:17703]
It seems OK in Firefox (it is not possible to go to the line).
Thanks

Bernadette

RE: mathml mathjax [ Reply ]
By: joke evers on 2016-10-24 17:08
[forum:17702]

wims_mathml.y (15) downloads
Since this is clearly a browser bug in safari, I suggest we return to the "textarea trick".

The attached file (wims_mathml.y) disables the "enter" on the textarea using :

The argument of Bernadette should now be by-passed.
(ofcourse we will return to the input as soon as safari-webkit implementation is up to standards (...)

Please test this on your systems and report back !


RE: mathml mathjax [ Reply ]
By: joke evers on 2016-10-19 10:42
[forum:17700]
Ok !
May be you should add the fact that a textarea (a per default closed html object) has no problems.




RE: mathml mathjax [ Reply ]
By: Olivier Bado on 2016-10-19 09:48
[forum:17699]
If you're interested in following, the issue has been posted in MathJax Repository here :

https://github.com/mathjax/MathJax/issues/1644

RE: mathml mathjax [ Reply ]
By: Olivier Bado on 2016-10-18 14:02
[forum:17688]
The problem is still present in Safari 10. + MathJax v. 2.7.0

I tested different solutions with no satisfying results, such as :

* changing WIMS Doctype from xhtml to HTML 5.0 has no impact.
* changing annotation-xml encoding :
- "text/html" or "application/xhtml+xml" works on FF, but not on Safari
- all other encoding tested (even no encoding) works on Safari, but not on FF.
=> All the things that works on safari does not work on Firefox, and all the things that works on Firefox does not work on Safari.

==>I think we can’t resolve this from our side : it’s a problem to be resolved by MathJax.

RE: mathml mathjax [ Reply ]
By: joke evers on 2016-10-18 07:06
[forum:17687]

mml.html (60) downloads
to be more clear:
Mathjax (< v2.6)+inputfields gave the exact same javascript error message.
on all (!!!) Mathjax enabled browsers
We fixed this by using a "textarea" (strangely mathjax had no problems with it...)

As of Mathjax v2.6 we tested again and the known bug seemed to be fixed in Mathjax.
tested:
windows : IE | chrome | opera
linux: gecko | opera | chrome
freebsd: gecko | opera | chrome

Therefor (and because of the "enter in textarea" problems) we reverted to the
usage of a real inputfield in wims_mathml

Someone with the "safari problem" should test the older wims_mathml-textarea-version

The usage of the tag <script type="math/mml"> is not possible : native mathml browsers
will silently ignore this unkown script type...



mathml mathjax [ Reply ]
By: Bernadette Perrin-Riou on 2016-10-18 06:22
[forum:17686]
I try here to make a resume of some discussions by mail on mathml and mathjax.

1. Recently, Olivier has desactivated native mathml for safari because some formulas
are not correctly written (overline ? sqrt ?). So Mathjax is used with safari.

2. We have not seen that the inputfields are not correctly parsed by Mathjax in safari
(no problems have been seen in chrome, chromium as I know):
error : something as
Error parsing MathML: error on line 12 at column 18: Opening and ending tag mismatch: input line 0 and annotation-xml
See for example
http://wimstest1.di.u-psud.fr/wims/wims.cgi?module=H5/geometry/oefgeoplan.fr&cmd=new&exo=ruche

3. There is some discussions in mathjax (2014)
http://stackoverflow.com/questions/21171107/mathjax-rendering-html-input-fields-in-mathml
Olivier remarks that they proposed to add some
<script type="math/mml"> ....</script>
But Joke has tested, it is not good for native mathml.

4. Before version 2.6 of mathjax, Joke had seen the problem and there was no inputfield
but textarea. The problem seems to be solved with version 2.6 (in wims, it is now 2.6.1, it seems to me, update by compile --mathjax )
She proposed to come back to textarea. The problem with textarea is that pupils using the enter button don't see that
what they are writing a second line, so it is not good in general). So I am not favorable
to do that for one navigator.

5. For the formulas not well written, it is always possible to go back to images.
There is no alternative to embed inputfields in mathematical formulas
(much prettier than the old tables).
So on my servors, I will keep mathml with safari if we don't find a solution.
and I am a little reserved to put in a release the modification (safari with mathjax)
as it completely bugs some exercises.

6. There seems to have "MathJax v2.7 now available" October 14, 2016 !!
We have to test it for that and for everything else !
It will be good that someone do a list of significative wims pages where mathml
is used.
That is the political reflexion I have this morning !

Bernadette