mozilla :: #mathml

15 May 2014
02:08nattokiraikarl: have you done much work in nsMathMLChar.cpp?
02:08karlabout 7 years ago
02:08nattokiraiso fredw is the main dev on that file?
02:09nattokiraicurrently it's *really* hard to untangle all the fontlist manipulation being done there...
02:10karlfred has removed quite a bit
02:11karlSetFontFamily hasn't seen much improvement though
02:11karloh, yes, it has, but it is still clumsy
02:11karlmost of the time it should be a no-op
02:12nattokiraifor me, the problem is that font names are being passed around all over the place
02:12nattokiraithere's no clear distinction between "list of fonts" and "a single font"
02:13karlyes, that derives from the nsIRenderingContext days
02:13nattokiraisometimes you end up with a situation where the fallback list has been inserted *twice* into the list of names
02:13karli wasn't aware of that
02:14nattokiraii just dumped out the params coming in to SetFontFamily
02:14nattokiraiand ran the MathML reftests
02:14nattokirai300 test ===> 132,000 calls to SetFontFamily (!!!!)
02:16nattokiraiseems like you should be able to use a *single* font group
02:16karlthere is one list in mathml.css for the CSS font-family and font.mathfont-family inserts fallback fonts,
02:17karland, yes, with those, it should be possible to make a single font group
02:17karlassuming apis to measure using each font
02:17karlwhich exist now, i assume
02:18nattokiraiit's not iterating over available fonts in the font group
02:18nattokiraiit's iterating over names in the fontlist (i.e. style font-family + fallbacks)
02:18karlyes, perhaps the api that is missing is making a textrun from a font
02:18karla single font
02:19nattokiraiwell, making a gfxFontGroup with a single gfxFont is sort of what you mean i think
02:20nattokiraibut, yeah, something like that
02:20karlwe do shaping with a single font
02:23nattokiraineed to figure out how to do it relying on gfxFontGroup to do basic matching functions
02:24nattokirairather than creating new gfxFontGroups all over the place
02:24nattokiraiok, i'll try and connect with fredw
02:24nattokiraiwhen is he usually online?
02:24karlhe's in paris, i think
02:25nattokiraiis his work sponsored by somebody?
02:25nattokiraihe's doing a lot of coding
02:25karlhe ran a crowdsourcing project, but that ended a few weeks ago
02:27nattokiraido he works paris daytime hours usually? or more during the evening?
02:27nattokiraii guess you usually chat with him during your afternoon or evening?
02:27karli don't think he has a new job yet; i see him active in reasonably early morning paris time
02:28nattokiraiok, that's good for me ;)
11:45physikerweltwill there be a meeting here at 2pm (in 15minutes)
11:46ranierephysikerwelt: hi
11:47ranierephysikerwelt: AFAIK the meeting is schedule to 1pm UTC (in 1:15)
11:50raniereraniere: did I mess with google calendar?
11:55fredwraniere: the google calendar says in 5 minutes for me
11:57ranierefredw: I will try to find what happe later
11:58ranierefredw: fscholz: jkitch: KhaledHosny: maxli: physikerwelt: teoli: Could we start the meeting in ~3 minutes?
11:59physikerweltI'll have to leave in 45 minutes from to now to catch a flight
12:01ranierephysikerwelt: Hi everyone, I'm starting our May meeting.
12:02ranieresorry to not run it last week since I had attend an foss event:
12:02ranierethe topics for this meeting are at
12:03ranierephysikerwelt: there has a topic that you want to discuss before your go out?
12:04physikerweltI just check the list
12:05physikerweltno I'm fine
12:05fredwphysikerwelt: what's the progress with the Mediawiki database review?
12:05physikerweltno progress
12:05physikerweltand even with the trival latexml change that uses now a latexml 0.8 server no progress
12:06fredwphysikerwelt: did you find someone from MediaWiki to review?
12:06fredwTheDJ is not from MediaWiki, but perhaps he knows about the database?
12:06physikerweltthe only good news is that mathoid0.2.4 was released wich now works with ubuntu 14 LTS out of the box
12:07fredwphysikerwelt: and also there is a web service I think
12:07physikerweltthere is springle from WMF but I'm not sure if he is still interested in doing a review
12:08fredwphysikerwelt: I can try to do a review again, but last time I check the change was big
12:08physikerwelt could be +2 without risk
12:08physikerweltfredw: if you can suggest how to break it down this might help as well
12:09fredwyes, I think so. but I think TheDJ wanted to review it
12:09fredwphysikerwelt: I'll need to understand exactly the change first. I'm not even sure it's possible to break it, though
12:10raniereI and fredw will be happy to have feedbacks for
12:10physikerwelt@fredw: however, I don't want to disturb your mathml meeting
12:11ranierephysikerwelt: have news from MathML at Wikipedia was nice. thanks
12:11raniereMoving to "Web developments"
12:11fredwIt's for the CICM in Portugal in July
12:11fredwraniere & I plan to go there
12:11fredwI think physikerwelt too
12:12fredwHe has a paper
12:12physikerwelthcohl will be there as well... maybe he will join this meeting later on
12:13fredwand probably many other MathML people will be there
12:13fredwphysikerwelt: Who is hcohl?
12:13physikerweltHoward S. Cohl
12:14raniereThe MathML CSS was merged into MDN.
12:14fredwOK. There will be many other people from LaTeXML etc
12:14raniereA new version of LaTeXML was released and you can help testing it.
12:15fredwespecially on Windows
12:15physikerweltthere is also a with a master student about Mathematical Language Processing that will presented either by me or by the master student who did the implementation (if we find some funding)
12:15fredwBTW, to everybody: feel free to add your name in
12:16ranierefredw: could you talk about your work integrating LaTeXML into LyX?
12:16raniereLyX is a WYSIWYG editor "for LaTeX"
12:16fredwLyX is a WYSIWYG LaTeX editor
12:17fredwI've written a patch to allow HTML5, HTML4 (with PNG images for math) and EPUB3 export via LaTeXML
12:17fredwhowever, I found a bug in LaTeXML, so it's not usable now
12:18fredwI've also done some experiments to use LaTeXML with TeXworks:
12:18fredwand I've created a tutorial for Windows users:
12:18fredwSo the idea is to help people to use LaTeXML, as I got feedback about the difficulty to install and run it
12:19ranierefredw: nice work
12:19fredwLaTeXML 0.8 now has many packages to help installation (Debian, CPAN etc)
12:20raniereMathJax Native MathML, fredw addon, need help with localization
12:20fredwI've improved the add-on with some options in the preferences
12:20fredwthis is what needs to be localized
12:21ranierefredw: I will send a pull request with portuguese localization
12:21fredwNow when for pages that already contain MathML input, MathJax does not try to process them at all. So it's more efficient
12:21fredwOK thanks
12:22physikerwelt^^ MathJax Native MathML can you post a link
12:22ranierefredw: what MathJax do for browsers without MathML support?
12:22fredwIt's an add-on for Firefox, so the MathML support is always there :-)
12:23fredw(I meant, the add-on forces MathJax not to process the MathML input)
12:23fredwThere is a version for Safari too (with less options)
12:24fredwI got a message from Apple that the add-on is accepted, but it's not on their gallery yet
12:24physikerweltthat's a big sucess as well
12:24physikerwelthave to leave cu
12:24fredwOne of the idea is that these add-ons can be used to expose the MathML for accessibility tools (e.g. VoiceOver)
12:25fredwphysikerwelt: ok, bye
12:25ranierephysikerwelt: bye
12:25fredwI'll take a look at the database change after the meeting
12:26ranierefredw: did you talk with Jonathan Wei?
12:26fredwraniere: no
12:27raniereJonathan Wei gave a talk about MathML accessibility:
12:28raniereI had invite him to the meeting but he couldn't attend
12:28ranierefredw: some comments about TeXZilla.toImage ?
12:28fredwIt's a new function in the dev version of TeXZilla
12:29fredwit generates an SVG image with a MathML formula inside, for use in Canvas
12:29fredwand one of the idea is to generate 3D schemas with WebGL:
12:30raniereany other topic about "Web developments"?
12:30fredwnot for me
12:30raniereMoving to "Gecko developments"
12:31ranierelots of bugs was close in the last couple of days. Thanks very much for all that make this possible, specially fredw.
12:31fredwThere is LaTeX-to-MathML in Thunderbird 31 (I think it was already mentioned in the previous meeting)
12:32fredwSo the most important change is the OpenType MATH table in gecko 31
12:32ranierefredw: I don't know what bug need comments. feel free to talk about any one.
12:32raniereanujagarwal464: ^
12:32fredwPeople can try an aurora build with
12:32fredwor to test
12:33fredwat the moment the biggest problem is for italic variables as OpenType MATH fonts don't have italic style
12:33fredwand one should use the characters in the Math Alpha Num unicode block
12:33fredwSo I'm currently working on jkitch's patch for mathvariant
12:34fredwI expect it will land soon and I plan to make a patch in a couple of days for Aurora
12:34fredwso that it is available in Firefox 31
12:34fredwFirefox 31 is the ESR, so it's good if we can start suggesting people to move to Open Type MATH fonts
12:35fredwand ask distrib maintainers, font designers etc to do the update
12:36fredwSo except fonts, I think the other change listed here is support for the phasorangle notation
12:37fredwActually, it was mainly done by Branko Krznaric and the tests by Anuj & Aniket
12:37fredwThis is introduced in the recent MathML3 revision and will be in MathJax 2.4
12:38fredwSo atm I'm still working on the font stuff
12:39fredw"drawing failure with stretchy horizontal parenthesis when no fonts are installed" has been fixed by anujagarwal464, I think. I should be moved to the "closed bugs" category
12:39anujagarwal464fredw: thanks for help
12:39fredwThe other important change is by jkitch, which fixes an old regression (that I introduced) with stretchy char
12:40fredwanujagarwal464: you're welcome
12:41fredwSo I think this essentially gives the overview of the current patches. Most of my work is on OpenType Math fonts, see the bug for the details
12:41fredwraniere: so we can move the Documentation section?
12:41ranierefredw: yes
12:42ranierefredw: teoli: could you talk a little about documentation updates?
12:42anujagarwal464fredw: yes
12:42fredwraniere: actually I did the documentation
12:42fredwthere were many "dev-doc-needed" bugs
12:43ranierefredw: sorry
12:43fredwwith all the changes in Firefox 28 and 29
12:43fredwso I tried to document all of them
12:44fredwNow there is only the phasorangle change to document:
12:45fredwSo if someone wants to do that, feel free to update the "Firefox 32 for developers" page and the menclose pages on MDN
12:48ranierefredw: any other topic about documentation?
12:48ranierefscholz: teoli: ^
12:48fredwThere is "More convenient way to include the mathml.css/mathjax fallbacks: just one script in the header"
12:49fredwNow you just need one js line for CSS and MathJax fallback and it will automatically detect if MathML is supported
12:49fredwthis is inspired from raniere's work on MDN
12:50raniereWG9s: welcome
12:50fredwand the OpenType MATH fonts, I already mentioned them before
12:50fredwWG9s: Hi
12:51WG9sHi Fred!
12:51fscholzOn the doc status pages, we now display a better overview table of the translation work, Besides that, I don't have more to report
12:51fredwWe've started the meeting one hour ago
12:51fredwfscholz: sometimes the translation status was not updated. Did you fix that?
12:52fscholzfredw: Yeah, there data gets refreshed every day now. There is still one glitch to fix, but it should be better already.
12:52fredwfscholz: ok, cool
12:52raniereWG9s: we start the meeting based on Google Calendar clock (maybe I use it in the wrong way and mess a little the time of the meeting)
12:53ranierefredw: could you point me where I can learn more about Glyph library used by gecko?
12:54fredwwhat do you mean?
12:54fredwthe graphics stuff is in gfx/
12:54WG9sraniere: it's not you. I was late.
12:55fredwBut Karl Tomlinson or Jonathan Kew know this better than me (although I got the opportunity to work on this part recently ;-)
12:55ranierefredw: in the past when I try to hack gecko I was lost because of the graphics stuffs and don't know the better way to learn it
12:56fredwYes, I think the best is to ask the appropriate person who is familiar with the code
12:56fredwso for the MathML code, I can help :-)
12:56fredwbut for other parts, it's better to ask someone else
12:57fredwit really depends on what you wanted to do
12:57raniereok. I will try again and ask you for help.
12:57ranierecould we move to "Miscellaneous"?
12:57fredwBTW, for font installation (if that's what you thought about) I wrote a patch
12:58ranierefredw: =)
12:58fredwAnd for the detection, I refreshed Jonathan Kew patch (I think that's what Karl suggested you to do):
12:59fredwI believe will be used for languages & math and for android & Linux.
13:00fredw is for Android
13:00raniereMy GSoC project was accepted and it start next Monday. I create a metabug to keep track of it:
13:01fredwcool. I'm already cc'ed ;-)
13:01raniereand updates will be post at (RSS:
13:01fredwPerhaps you want to say a word on the current progress?
13:02raniereuntil know I only do very small and basic hacks in many fronts
13:03ranierein the next couple of months I need to extend and merge this hacks
13:04ranierethis weekend I want to hack to provide a online demo of the keyboard like
13:05raniereAnd last week I attend a FOSS event here in Brazil to give a talk about MathML:
13:06ranierein my talk I show almost all of the demos create by fredw for "Mathematics in ebooks" project
13:06ranierefredw: the public loved your demos!
13:06fredwraniere: I think Seamonkey and LyX could soon be added to WYSIWYG authoring mathml tools for students
13:06fredwraniere: thanks
13:07ranierethe question that always pop up was "What tool can I use to do that?"
13:07fredwyes, that's why we need better tools
13:07fredwwe know that i'ts possible to do that by hand, but let's write tools to do that
13:08fredwThat's one of the idea of the Math suite in our MathUI presentation
13:08raniereI discovery that many cities here in Brazil have money to buy authoring tools but they don't want to use the money to support building the tools =(
13:09fredwraniere: is it proprietary tools? I thought all the administration in Brazil moved to Linux, right?
13:10ranierefredw: only part of the administration. We have a law saying that we need to prefer free/open software but if no such one exist we can use proprietary one
13:10fredwFor those who missed my "Mathematics in ebooks project", the demos are
13:11fredwraniere: ok, I see.
13:11raniereI believe that we cover all the topics
13:12ranierefredw: anujagarwal464: fscholz: jkitch: KhaledHosny: maxli: teoli: WG9s: any extra topic?
13:12jkitchnot from me
13:12WG9snothing here.
13:13fscholznot here. thank you very much!
13:13fredwraniere: did you update the MathUI paper?
13:14ranierefredw: nope I will do it after lunch
13:14raniereThanks all for your time.
13:14raniereHope to see all in our next meeting at 06/12.
13:15fredwraniere: finally did you get a reply from the conf funding
13:15ranierefredw: not yet
13:15ranierestill waiting
13:16ranierefredw: sorry for the long delay to update the MathUI paper, last couple of days was very busy.
13:16fredwok, np
13:16fredwanyway they moved the deadline
13:17ranierefredw: should I create a new subsection for the keyboard or only add a reference?
13:17fredwI think you already created a subsection?
13:17fredwso you only need to reference it now
13:17raniereyes, I did but not sure if it was the best option.
13:18fredwOK, perhaps you can only move the content where you mention the keyboard (that's what I thought you'd do)
13:19raniereI'm having a small problem with the citation that I add since "make pdf" won't insert it properly
13:21ranierefredw: could you give me a little help ^
13:22fredwI let you do all your changes and will check the paper again afterwards
13:22ranierefredw: ok
16 May 2014
No messages
Last message: 3 years and 102 days ago