mozilla :: #mathml

14 Aug 2014
14:15djaniketsteris there a specific source tree of firefox we use for mathml patches?
14:37ranieredjaniketster: all the mathml patches goes directly into firefox's source tree
14:38ranieredjaniketster: if you are searching for where the mathml code is, big part of it is at firefox_root/layout/mathml
19:24djaniketsterdo we use specific mozilla source tree for landing the patches? or we can use any of the available?
19:25fredwdjaniketster: development is on
19:27djaniketsterfredw: okay. thanks!
19:44fredwhi raniere. The meeting is in 20in, right?
19:53physikerwelt@fredw: Thanks for the reminder... I had almost missed the meeting
19:57fredwphysikerwelt: actually, I'm not sure what time the meeting is. raniere, are you here?
20:06ranierehi everyone. sorry for the delay
20:07fredwIf you want to go video meeting:
20:07ranierekarl: djaniketster: fscholz: maxli: physikerwelt: teoli:
20:08raniereI and fredw want to try WebRTC for our meetings. could you join us?
20:10ranierefredw: more 5 minutes?
20:11physikerweltCurrently I have no audio/video availible
20:12ranierephysikerwelt: maybe we back to IRC
20:12ranierefredw: ^
20:13ranierewe are going to delay the WebRTC test. if anyone want to join the video later we be happy to talk
20:13raniere1) Main topics (general things to discuss first...)
20:14ranierefredw: do you want to talk about MathUI and others conference funding?
20:15fredwso I think Mozilla replied on the bug for funding request that fscholz ask us to open
20:15fredwbut they said they won't fund it but will try a new system
20:15fredwnot sure if anyone at Mozilla has more info?
20:17raniereAbout MozFest, Kaitlin had request people to send proposals
20:17ranieremaybe someone had something that could feet at MozFest (Education + Technology)
20:18ranierephysikerwelt: do you know people from Wikimedia that are working to improve Wikibooks?
20:19ranierefredw: do you have something that you can show?
20:19physikerweltraniere: no
20:19fredwraniere: no
20:20raniere2) Web development and authoring tools
20:21ranierephysikerwelt: could you give some updates abotu MediaWiki?
20:21physikerweltMathoid has been deployed on beta-labs
20:22physikerweltunfortunately all MathML tags are filterd out by HTML Tidy that does not accept MathML
20:22physikerweltI'm working on a patch for that right now
20:23physikerweltthat means after has been solved MathML is ready to move to production
20:23physikerweltof course the move to production is a step that requires additional attention
20:24physikerweltoperators have to assign hardware resources for mathoid
20:24ranierehuxuan: welcome to our meeting and I hope that you enjoy it. If you like you can present yourself
20:24physikerweltand if more bugs are being found during the test phase of mathoid in beta-labs they have to be fixed as well
20:24fredwphysikerwelt: what's the plan for Image/SVG/MathML?
20:25huxuanSorry for coming late. My name is huxuan, and you can call me Sean in English. I am from Beijing, China and currently working as an intern in Google Switzerland.
20:25fredwwhat will you deliver to people (and anonymous users)?
20:25physikerweltfredw: can you be more specific
20:25fredwBecause at the moment SVG is used for all but Firefox, right. Will that be the case in production.
20:26fredwOr do you plan to continue to serve PNG images by default first
20:26physikerweltin a first phase PNG will stay the default
20:26physikerweltWe need to be extremly conservative here
20:27fredwyou mean the "PNG only" or PNG+SVG+MathML (SVG and MathML hidden)
20:27physikerweltfirst the PNG+SVG+MathML will be availible to registerd users only
20:28physikerweltat some point in time we will change the default user option
20:28fredwphysikerwelt: I think it would be best to give PNG+SVG+MathML to everybody but hides SVG/MathML by default.
20:28physikerweltit's just a bit in the config
20:28fredwSo that people can write browser extensions to show the SVG/MathML if they want
20:28physikerweltfredw: I think that would be a good second step
20:29physikerweltOnce we have mathoid in production we can submit patches to the code which are deployed once a week to production
20:30ranierephysikerwelt: could you send a list with links to wiki pages with many math expressions to us for test?
20:30fredwIs it already available on betalab?
20:30huxuanI have some
20:31physikerwelthuxan: we could copy those pages to betalabs
20:31ranierehuxuan: can you disclose what are you working on at Google?
20:32fredwraniere: what is the link to the beta-lab
20:32fredwphysikerwelt I meant
20:32fredwphysikerwelt: so IIUC, it only works for registered users and MathML is filtered out by tidy right now?
20:33huxuanraniere: I am in YouTube team, currently working on some features for creators.
20:33ranierehuxuan: =)
20:33fredwhuxuan: you're no longer working on the math search engine?
20:33ranierephysikerwelt: thanks for the update and the link for beta-lab
20:34fredwphysikerwelt: do you have a user account to test with?
20:34ranierefredw: do you mind talk about the others issues at web development?
20:34physikerweltI think you have to create one
20:34huxuanfredw: I am still working on that as it's the topic for my master thesis.
20:35huxuanJust working as an intern for conversion opportunitity.
20:35huxuanI should graduate next year so it's job season currently.
20:35ranierehuxuan: maybe I will request some help for my crease idea of client-side javascript math search engine
20:36fredwphysikerwelt: is there a math page on betalab?
20:36huxuanraniere: It's my honor if there is anything I can help.
20:37physikerweltfredw: I have not created one... a first tests points to
20:37physikerweltjust now I created
20:39fredwphysikerwelt: thanks, I see that indeed the MathML is filtered out
20:40physikerweltI'll try to write the test requested by hashar
20:40huxuanseems we are talking about the resource for testing?
20:40physikerweltI updated
20:40physikerwelthowever the link to is still a little bit hidden
20:43physikerweltI think I'm done with the update from my side
20:43physikerwelthuxuan: could you send a reference to your work on mathsearch
20:44fredwphysikerwelt: ok, thanks. I'll try to +2 the patch to workaround the tidy bug tomorrow, if nobody complains
20:45physikerwelthuxuan: Are you aware of ? If you are interested in participating... I'll invite you once the second round starts
20:45physikerweltfredw: It's a core patch... so don't worry about that someone will merge it
20:45huxuanphysikerwelt: this is the second version.
20:45fredwphysikerwelt: ok
20:47huxuanAnd for the ntcir competition, actually I have took part in the Math-2 Task.
20:47ranierefredw: do you mind talk about the others issues at web development?
20:48huxuanBut I don't think we are doing well cause the limited time (though there is already an extension for the deadline), just busy with some other staffs.
20:48fredwthere is other latexml stuff, but latexml people are not here
20:48fredwBut if someone is interested on working on LaTeXML, would be a good one
20:49physikerwelthuxuan: I think you are right... but I'm not authorized to publish the results
20:50fredwThere was also discussion on creating a Latin Modern Math fork to fix issues (will be useful for mediawiki for example) as Cambria Math does not work very well on Windows
20:50physikerweltfredw: I think you could email to deyan
20:50fredwphysikerwelt: which one
20:50physikerweltmaybe he is interested in joining to discuss that
20:51physikerweltfredw: I was referring to
20:51fredwphysikerwelt: I think we invite him and bruce and michael etc
20:51fredwbut they don't come :-(
20:51fredwAnyway, I think this is not a priority for them but maybe someone in Mozilla MathML is interested to give a try
20:52huxuanphysikerwelt: Yeah, it's OK, I don't suppose a smei-finished system can have some too good results.
20:52physikerweltfredw: can you past the link to this chat?
20:53fredwphysikerwelt: which link?
20:53ranierephysikerwelt: you can filter the date by the URL
20:54physikerweltfredw: I was looking for the full url of the moziall irc server
20:54ranierephysikerwelt: ?
20:54physikerweltraniere: thanks
20:55ranierecan we move to 3) Gecko development ?
20:55fredwraniere: ok
20:55fredwI think most of the work has been done by James Kitchener, but he is sleeping
20:55fredwSo there was a bug reported by Bruce and physikerwelt and others
20:56fredwabout the fact that MathML is slow on large page after the mathvariant implementation
20:56fredwJames is working on that
20:56ranierethe first big issue that people had work at the past month was 10143358 that slow down geck when reading mathml
20:57fredwkarl suggested that it was due to a MathML change in the font cache
20:57raniereI want to set a server to benchmark it but I couldn't find informations on how do the banchmark
20:57fredwso James tried another patch
20:58ranierehi Deyan. be welcome to our meeting.
20:58Deyangreetings MathML folks :)
20:58fredwphysikerwelt: you can try
20:58fredwand see if you get better results
20:58fredwhi Deyan
20:59huxuanWelcome! Deyan!
20:59raniereDeyan: if you like to introduce your self, please do so. we are talking about the "regression" at gecko that it make slow when rendering mathml
21:00fredwDeyan: you can also try if you have audio or video
21:00DeyanOh, where are my manners. For this discussion, I am a LaTeXML dev and a Content MathML enthusiast (Twitter handle: )
21:01ranierefredw: do you know how can I get the time that gecko use to render some page?
21:02fredwraniere: nope :-(
21:02* raniere going to keep searching
21:03fredwDeyan: do you want to talk about any recent progress on LaTeXML?
21:03DeyanOk, I caught up with the backlog. Sorry I hadn't managed to join the conversations before, and thanks to Moritz for prodding me :)
21:04huxuanraniere: The 'Network' tab in 'Developer Tools' should have some time related statistics. but I am not sure whether that is what you want
21:04fredwI just mentioned in case some volunteer wants to work on that. That should be easy, but will probably require some mentoring from LaTeXML devs
21:04DeyanI could give you a brief update, sure, though it is tangential to MathML as a topic
21:05fredwDeyan: well, we mostly spoke to MediaWiki and Math search developments, and not really Mozilla so far :-)
21:05ranierehuxuan: AFAIK the network tab keep track of time to download files (CSS, JS, JPG, ...) that can block the final renderization
21:06ranierefor mathml I won't need any CSS, JS, ... so I belive that it won't help very much
21:06DeyanSo, recent LaTeXML development is driven almost entirely by Bruce, and is targeting a near minor 0.8.1 release.
21:06DeyanThat will have mostly stability patches and minor features
21:06huxuanOr maybe we can add some debug parameters and then show the rendering time?
21:07huxuanActually I am not familiar with Gecko development, just share some potential ideas.
21:07ranierehuxuan: no problem. thanks for the suggestions
21:07DeyanI have invested a lot of time into rewriting core bits of LaTeXML into C, en route to pursuing long term performance boosts
21:07fredwraniere: huxuan: I think you can do some C++ profiling
21:08fredwDeyan: ok good news
21:09ranierefredw: sure by that. the only problem is that I don't know the function that Firefox use to render the pages.
21:09raniereDeyan: good news.
21:11karlraniere: i would look at what talos uses
21:11physikerweltDeyan: By the way are you still working on the improvement of content MathML?
21:12karlraniere: but i don't know whether it is fancier than measuring page load time
21:12ranierekarl: thanks for the tip. could you send me a link for talos?
21:12karlraniere: there is mozAfterPaint, but i don't know how reliable that is, and mozPaintCount
21:13Deyanphysikerwelt: I have the desire, but not the time :/
21:13physikerweltI'd like to evaluate the idea of different possible meaning to overcome the ambigious entries in LaTeXML content MathML output
21:14DeyanI have mostly pushed content mathml politically, sent some suggestions to WolframFunctions on how to start improving their cmml
21:14Deyansince it is quite junky
21:14karlraniere: but that interface is painfully slow, so hg clone
21:15ranierekarl: thanks for the link
21:15Deyanphysikerwelt: you mean having ContentMathML that represents ambiguous parses put together?
21:16DeyanI have a version of that, but it requires a lot of work to make more complete
21:16physikerweltDeyan: I don't see a solution how to decide that for example superscript really means power?
21:16DeyanRight, that's AI
21:18Deyanhence the "requires a lot more work" part
21:18physikerweltYes, but for example for Wikipedia we could assume with a very high probability that it means power unless the article is in a certain namespace
21:19physikerwelta human would have to decide in the end if ^ really means power or not
21:19DeyanI am tempted to go hunt for counter examples :)
21:19Deyanif you have the luxury of asking a human, that is wonderful
21:19Deyanbest way to create content mathml
21:19physikerweltBut we need to have a starting point that does a relativly good job
21:20Deyanbut there are many other problems ahead, if you want to cover all of math (= all of arXiv / wikipedia)
21:20Deyanthe content dictionaries that define csymbols are too few and cover mostly highschool and undergrad math
21:21physikerweltI think we need to start somewhere... restricting to K12 does not seem a very bad idea
21:21physikerweltby the way your counter example
21:21DeyanWolframFunctions and DLMF, both of which have content mathml, don't have valid csymbols for their domain-specific symbols
21:22physikerweltI talked to Bruce yesterday about that... the CD is on his todo list
21:22Deyanphysikerwelt: excellent, index variables are a good counter-example :)
21:22raniereI believe that I talk with fredw about something similar for TeXZilla
21:23fredwraniere: well, I don't think TeXZilla does any content MathML ;-)
21:24physikerweltDeyan: As far as I know you are the only person who ever had another way of generating contentMathML I would love to see someone continuing with this work
21:25Deyanphysikerwelt: thanks. Michael also has a continued interest in the ambiguous representations, so even if I don't get the time, KWARC will take it up eventually
21:26ranierefredw: sorry, I missed the content mathml.
21:27raniereDeyan: Thanks for the updates
21:27raniereDeyan: physikerwelt: can we move to the next topic of the meeting?
21:27physikerweltHoward proposed to do a student project for this issue... I think there is a high initial investment to get started with even simple things
21:28physikerweltraniere: sure..
21:28ranierethe last couple of weeks I was working to finishing my Google Summer of Code project
21:29raniereyou can find the results of it at (there are some gifs =))
21:29raniereand it is available at marketplace.
21:29fredwraniere: congrats
21:29fredwwhen is the end of the gsoc project?
21:29ranierefredw: this weekend
21:30huxuanraniere: big cong!
21:30Deyanraniere: cool project! :)
21:30raniereif anyone want to send me comments based on the previous link that I post I will be very happy
21:30fredwraniere: we need FirefoxOS to test
21:32ranierefredw: thats true. but I will be happy with comments like "What about add this feature" or "What about do this that way" even if you didn't test it with a real device.
21:32huxuanraniere: Yeah, I also wonder how can we get a firefox os mobile?
21:32fredwraniere: apparently some people didn't understand the use
21:32fredwperhaps you should mention the TeXZilla app
21:33fredwor the LaTeXML service
21:33ranierehuxuan: there are some devices available:
21:33ranierefredw: sure.
21:34Deyanraniere: If you have the keyboard, what apps do you use it in? Web sites like MediaWiki? Since phones don't have text messages / emails / files that support TeX
21:34huxuanraniere: Thanks for the link.
21:35ranierefredw: karl: Deyan: fscholz: huxuan: maxli: nattokirai: physikerwelt: teoli: That's all. Thanks for joining the meeting and hope to see you next month.
21:35physikerweltraniere: Do you have an option to enter a list of frequent terms... I created one based on enwiki (
21:35raniereDeyan: MediaWiki.
21:35fredwDeyan: yes, to edit MediaWiki or any places where you enter LaTeX
21:36fredwone of the goal of TeXZilla was to make such FirefoxOS apps possible. But at the moment there are not many of them
21:36huxuanraniere: I think that will works fine with search engine supporting LaTeX query. :-)
21:36fredwhuxuan: yes ;-)
21:36raniereDeyan: some times you could need to send a small tex expression for someone at email as plain text
21:36fredwSo if people want to try
21:37fredwbefore we end the meeting
21:37fredw... then I'll go to bed
21:37Deyanraniere: fredw: that makes sense, thanks
21:38fredwraniere: ?
21:39ranierefredw: sorry. power supply problem. forgot to connect my laptop at the power plugin
21:39fredwraniere: ok no problem works pretty nicely, impressive
21:40fredwDeyan: we almost were 4 at the same time
21:40raniereDeyan: could you back for a minute?
21:43raniereThanks you all again for this awesome meeting
21:43fredwok, thank you everybody
21:44fredwNext time I hope we will be more people on Webrtc
21:44huxuanGlad to see all you talented guys!
21:44fredwso that we can speak faster :-)
21:44ranierephysikerwelt: the English/... keyboard at firefox OS has a suggestion line
21:44Deyandisagree on speed, typing is much faster than speaking ;-)
21:44huxuanActually I am a new comer, wish I can involved more in the future.
21:44Deyanbut it's fun to have video; good night all, i will try to join again in the future
21:45fredwDeyan: well, we compared and you took too much time to speak with physikerwelt than between us
21:45huxuanHave a good night/day!
21:45raniereI didn't port it yet for the LaTeX keyboard because it will request more time to understand the code and because it will reduce the space that user has available to read what he/she is typing
21:46fredwgood bye !
21:46ranierebye every one.
21:46Deyanfredw: not sure I got your point, but goodbye
21:47physikerweltok goodby I really enjoyed this meeting
15 Aug 2014
No messages
Last message: 32 days and 22 hours ago