I am dipping my toe in MATH for the first time and finding the results somewhat curious. They key appears to be this statement:
"It generates either PNG images or simple HTML markup, depending on user preferences and the complexity of the expression."
Consider the formulas here:
http://en.wikipedia.org/wiki/Spherical_tokamak
Is there any hope that the PNG and HTML versions of things might be made to look more similar? The characters don't even look the same in some cases (kappa for instance) and this leads to VERY confusing output.
Maury
2011/1/19 Maury Markowitz maury.markowitz@gmail.com
I am dipping my toe in MATH for the first time and finding the results somewhat curious. They key appears to be this statement:
"It generates either PNG images or simple HTML markup, depending on user preferences and the complexity of the expression."
Consider the formulas here:
http://en.wikipedia.org/wiki/Spherical_tokamak
Is there any hope that the PNG and HTML versions of things might be made to look more similar? The characters don't even look the same in some cases (kappa for instance) and this leads to VERY confusing output.
You can force png rendering both by preferences and by code. But what's more interesting is the use of badly documented \scriptstyle TeX tag, which generates a much smaller and less "invasive" display of pngs: se this recent talk into en.source: http://en.wikisource.org/wiki/Wikisource:Scriptorium#Help.21_.28fractions_an...
Alex brollo
2011/1/19 Alex Brollo alex.brollo@gmail.com
2011/1/19 Maury Markowitz maury.markowitz@gmail.com
I am dipping my toe in MATH for the first time and finding the results
somewhat curious. They key appears to be this statement:
"It generates either PNG images or simple HTML markup, depending on user preferences and the complexity of the expression."
Consider the formulas here:
http://en.wikipedia.org/wiki/Spherical_tokamak
Is there any hope that the PNG and HTML versions of things might be made to look more similar? The characters don't even look the same in some cases (kappa for instance) and this leads to VERY confusing output.
You can force png rendering both by preferences and by code. But what's more interesting is the use of badly documented \scriptstyle TeX tag, which generates a much smaller and less "invasive" display of pngs: se this recent talk into en.source: http://en.wikisource.org/wiki/Wikisource:Scriptorium#Help.21_.28fractions_an...
Alex brollo
I added some \scriptstyle tags into inline expressions, and I discovered that they force png too, while giving a smaller (better in my opinion) display of tha formulae.
feel free to rollback! It's a test only.
Alex brollo
On Wed, Jan 19, 2011 at 8:54 AM, Alex Brollo alex.brollo@gmail.com wrote:
You can force png rendering both by preferences and by code. But what's more interesting is the use of badly documented \scriptstyle TeX tag, which generates a much smaller and less "invasive" display of pngs:
The use of scriptstyle to control font size is an ugly hack that makes many formulas less readable. For example, <math>2^{A_C},</math> and <math>2^{A}C,</math> become much harder to distinguish in scriptstyle.
In a custom installation, you can tweak the font size in images by changing the value of the "-D" parameter in line 8 of render.ml and recompiling texvc.
The ideal solution for Wikipedia would be to move to a system in which users with relatively modern browsers don't see images at all. There is already a candidate for that system: MathJax. This has extensive browser compatibility [1] and is actively maintained, with some big-name sponsors behind it [2]. The main difficulties enabling it on WIkipedia would be configuration and checking for any inconsistencies with texvc (so the main limitation is developer interest). On a custom install without a huge base of existing text, you could probably just use Extension:MathJax, although I haven't tried it.
- Carl
1: http://www.mathjax.org/resources/browser-compatibility/ 2: http://www.mathjax.org/sponsors/
Wow, thanks for the pointer Carl, MathJax is impressive.
Alex, your work is appreciated, but I'm not sure exactly what I'm seeing. Can you point me in the right direction to read up a bit more?
Maury
2011/1/19 Maury Markowitz maury.markowitz@gmail.com
Wow, thanks for the pointer Carl, MathJax is impressive.
Alex, your work is appreciated, but I'm not sure exactly what I'm seeing. Can you point me in the right direction to read up a bit more?
Don't care, throw away my suggestions,,,, and ask Carl. "Ubi major minor cessat!" :-) Alex
On Wed, Jan 19, 2011 at 11:56 PM, Carl (CBM) cbm.wikipedia@gmail.com wrote:
The ideal solution for Wikipedia would be to move to a system in which users with relatively modern browsers don't see images at all. There is already a candidate for that system: MathJax. This has extensive browser compatibility [1] and is actively maintained, with some big-name sponsors behind it [2]. The main difficulties enabling it on WIkipedia would be configuration and checking for any inconsistencies with texvc (so the main limitation is developer interest).
When I load their homepage, the formulas don't appear for about two seconds of 100% CPU usage, on Firefox 4b9. And that's for two small formulas. I'm not impressed. IMO, the correct way forward is to work on native MathML support -- Gecko and WebKit both support it these days, and Opera somewhat does too. I'm sure the support is a bit spotty, but if Wikipedia used it (even as an off-by-default option) that would surely drive a lot of progress. These days (with the deployment of HTML5 parsers) it can be embedded directly into HTML, it's not limited to XML.
On Fri, Jan 21, 2011 at 7:08 AM, Aryeh Gregor Simetrical+wikilist@gmail.com wrote:
When I load their homepage, the formulas don't appear for about two seconds of 100% CPU usage, on Firefox 4b9. And that's for two small formulas.
Works OK in Safari. WebKit perhaps?
I'm not impressed. IMO, the correct way forward is to work on native MathML support
I used to think that too. Then I looked at the examples on the wiki page on the issue. Although I find TeX rather opaque, a much worst issue is obscurity through verbosity, which not only makes the formula difficult to understand, but the entire source of the article too. That's why I don't use CITE either.
On Sun, Jan 23, 2011 at 4:24 PM, Maury Markowitz maury.markowitz@gmail.com wrote:
I used to think that too. Then I looked at the examples on the wiki page on the issue. Although I find TeX rather opaque, a much worst issue is obscurity through verbosity, which not only makes the formula difficult to understand, but the entire source of the article too. That's why I don't use CITE either.
We'd be talking about translating LaTeX input to MathML output automatically here -- no MathML input in the wikitext. It's certainly true that MathML is human-readable barely if at all, but it's more human-readable than PNG, which is what we output now. ;)
On Mon, Jan 24, 2011 at 8:09 AM, Aryeh Gregor Simetrical+wikilist@gmail.com wrote:
We'd be talking about translating LaTeX input to MathML output automatically here -- no MathML input in the wikitext.
Ahhh, I get it. And yes, that does make sense to me.
Aryeh Gregor wrote:
When I load their homepage, the formulas don't appear for about two seconds of 100% CPU usage, on Firefox 4b9. And that's for two small formulas. I'm not impressed. IMO, the correct way forward is to work on native MathML support -- Gecko and WebKit both support it these days, and Opera somewhat does too. I'm sure the support is a bit spotty, but if Wikipedia used it (even as an off-by-default option) that would surely drive a lot of progress. These days (with the deployment of HTML5 parsers) it can be embedded directly into HTML, it's not limited to XML.
Looking at http://www.mathjax.org/demos/tex-samples/ it may indeed take a couple of seconds to convert from TeX to the graphical view, but without 100% CPU usage or looking "blocked". I'm not using 49b but 3.6.12, though. I see a similar result in chromium. A disadvantage is that the showing the formula needs to reposition the content, instead of reserving the space in advance.
On Sun, Jan 23, 2011 at 7:44 AM, Platonides Platonides@gmail.com wrote:
Aryeh Gregor wrote:
When I load their homepage, the formulas don't appear for about two seconds of 100% CPU usage, on Firefox 4b9. And that's for two small formulas. I'm not impressed. IMO, the correct way forward is to work on native MathML support -- Gecko and WebKit both support it these days, and Opera somewhat does too. I'm sure the support is a bit spotty, but if Wikipedia used it (even as an off-by-default option) that would surely drive a lot of progress. These days (with the deployment of HTML5 parsers) it can be embedded directly into HTML, it's not limited to XML.
Looking at http://www.mathjax.org/demos/tex-samples/ it may indeed take a couple of seconds to convert from TeX to the graphical view, but without 100% CPU usage or looking "blocked". I'm not using 49b but 3.6.12, though. I see a similar result in chromium. A disadvantage is that the showing the formula needs to reposition the content, instead of reserving the space in advance.
Delurking to say that while I don't know if it's useful for us at all, Mathjax is getting lots of buzz in other settings (like publishing and the science library world); and also I just today came across this http://detexify.kirelabs.org/classify.html
It's not directly applicable but it is a fun usability idea for turning symbols into LaTeX (and by extension I can imagine symbols to markup, letters to unicode, etc.)
-- phoebe
wikitech-l@lists.wikimedia.org