mozilla :: #mathml

13 Feb 2014
10:39jkitchThe reftest dir-6.html is listed as "fails ==". Am I correct in saying that having this test pass is a good thing?
10:41fredwjkitch: yes, I think it should pass.
10:42fredwthe tests marked "fails" are for issues that have not been fixed yet
10:43jkitchGood My rowspacing work seems to have fixed it.
10:58fredwjkitch: that's great!
10:59fredwIIRC, the column spacing was too small in RTL
11:57jkitchin that case the test is broken in a different way. both tests now have the narrow spacing.
12:00fredwperhaps the test is not good. It would be interesting to compare with normal HTML tables when one sets direction: rtl on it
12:24jkitchno, it appears my default columnspacing code is faulty. Wide spacing appears when columnspacing is explicitly set to the default value.
12:30jkitchsilly typo. dir6 now works properly.
12:35fredwjkitch: ok good to know. (actually I'm not sure why the spacing was different in RTL)
13:26fredwraniere: did you finally figure out how to send the invitations with Google Calendar?
13:30ranierefredw: no with the link that I got
13:30raniereand I don't know the email from the folks that had to invite
13:30fredwraniere: what was it?
13:32fredwfscholz: do you know how to give edit permission to raniere for the Google Calendar?
13:33fscholzI need an email address to add him
13:35fscholzraniere: I added you
13:37fredwfscholz, raniere: I've edited the list of invited
13:37fredwraniere: check if you can edit the event, though
13:37ranierefredw: fscholz: thanks. yes. I can edit it
13:38ranierethere is something that I nedd to edit?
13:42fredwI just sent the invitations, so I think it's ok now.
13:43ranierefredw: thanks. I saved the list of email and add the description with the link to PAD's meeting
20:36WG9sfredw: good job no issues applying your queue to mozilla-central tip on my build today.
20:36fredwWG9s: ok, good to know.
20:37WG9sperhaps my heads up on the issues helped?
20:38fredwWG9s: yes, thanks. I was aware of the reftests issue since I wrote the patches that added them.
20:38WG9sI figured that one was kind of obvious anyway.
20:38fredwWG9s: however, I don't pull mozilla-central daily so I can't guarantee it will always apply
20:39fredwWG9s: are your builds ready, right now? or are they still building?
20:40WG9si do builds daily by pulling the mozilla-central tip and then applying the patches so I find out earlier that I would expect you would. I do it that way specifically to find such issues.
20:41fredwI'd like to figure out where the crash with comes from...
20:41WG9sso, when I tell you about issues it is not becuase i am comlaining. i ususally have gotten it to build just trying to let you know which patches need some love.
20:43fredwWG9s: yes, I know. Because your are building daily, I try not to push the changes when doing some large rewriting (like I did recently) so that's why I hadn't updated them for some time.
20:44fredwAlso, I try to always keep the "status" files to the patches I think you can include safely
20:44WG9sbut i also build daily by doing full source clobber of mozilla-central tip and then re-apply all the patches, so even different thatn jut rebulding.
20:45WG9si might find things the have merge issues doing it that way that would nto be merge issues if i took my tree adn merged in the tip.
20:46WG9sbut I think for non-landed patches the say I am uilding is beter because it identifies issues early.
20:48WG9sactually I don;t apply it to current tip i apply it to the changest the oficial nightly is built off of.
20:48fredwWG9s: yes, that's what I meant by "building daily"
20:48WG9ssi my build is supposed to be identical with offical nightly excpet for the identified patches.
20:48fredwok, I didn't know that (about the changeset)
20:49fredwWG9s: I thought you just pulled the latest mozilla-central
20:49fredwso in theory that means I might find issues before you :-)
20:49WG9sI habve this odd thing i run at 3M US pacific time that looks for bulds runnig that started after 3AM off mozilla-centra with the string nightly to see what chageset they used and it does my build on the smae cangeset.
20:50WG9ssso on any given day my builds are supposed to be the same as the officail nightlies with the additional patches.
20:52WG9sreson i do it that way is these are the builds i run so if i run into a repatabe crash means i can run the nightly adn submit the crash and then report a maningful bug.
20:53ranierefredw: I'm here. Just waiting some minutes to start our meeting.
20:53WG9sand if it crashes on my build and can not repeat on nightly then i knw it is related to the additional crashes.
20:54fredwraniere: ok. you are the boss today ;-)
20:57WG9sSo he should pick his 5 songs?
20:57WG9sand I knw you did not get that reference.
20:58fredwWG9s: yes, I didn't
20:59WG9shas to do with Sirius/XM E-Street Radio channel. they have this thing called be the boss where you mail them 5 bruce songs you want to here they call you and you record the intros to each adn you do your own half hour radio show.
21:04WG9sOK. I turned down the phone on my work meeting. it has beengoing on since 9AM and is very boring.
21:04WG9sit is 4PM here. so quite some time.
21:05raniereteoli: Can you please take a look at
21:05teoliraniere: sure
21:07WG9shmm what did those DCC messages mean?????
21:07raniereWG9s: sorry about that, there is a better way to notify all users?
21:08WG9sOH OK that was you?
21:08fredwraniere: I'm not sure it's a good time to take coffee for everybody...
21:09fscholzheh, true :)
21:09WG9sI might have a beer.
21:10fredwfscholz: the documentation status looks great. Is there something similar to track translation of articles?
21:10fscholzfredw: I am working on it.
21:10fredwoh, good :-)
21:10ranierefredw: do you mean the MDN MathML documentation?
21:10fscholzWe might have something in the near future.
21:11fscholzLets not skrew the agenda now
21:11fredwraniere: I meant for MDN in general
21:11fscholzReady to start?
21:11raniereWelcome everybody.
21:11raniereSome updates from our last meeting.
21:12raniere1) Now we have a log of the channel at
21:12WG9soh oh, so now we ned to be more carful of what we say?
21:13raniereWG9s: probably.
21:13fredwotherwise you'll end up in
21:14WG9sAnyway for those who acutally run the builds i provide on my website they are now actually built with tests enabled. also if you do an aobut:config the cahgeset reproted is the mozilla-central changeset upon wich the additianl aptches were applied.
21:14ranierefredw: can you give some info from the bug 941611 - Incorrect preferred width of italic characters
21:14WG9sadn shoudl be the same as that diays nightly.
21:14fscholzHow did you get add to that logging? interesting for other channels as well I think
21:14WG9sunless nightlies were respun adn i was not at a place where I could respin my builds.
21:15fredwfor the intrinsic width, James fixed the remaining issues on Mac/Windows
21:15ranierefscholz: some one that have OP privilegies on the channel should contact byron
21:16ranierefscholz: <>
21:16ranierefredw: nice =)
21:16fscholzraniere: thanks
21:17raniereCan we go to next topic (web developments)?
21:17fredwWG9s: so what does it change to have &quot;tests enabled&quot;. Do you run the tests?
21:18WG9sno but if you download the builds you can run tests.
21:18fredwWG9S: ok
21:18fredwraniere: sure
21:19fredwSo we now have various new Firefox add-ons
21:19ranierefredw: can you talk about all your new firefox add-on?
21:19WG9smy old server is used to do the builds was so slow that i had to build without the tests to get all the builds done by 1PM my time.
21:19WG9sthe new one can get them doe with tests by about 8:15 AM
21:19fredwThere is one add-on to force MathJax to use native MathML
21:20WG9sexcet on days i have to deal with merge conflict of course.
21:20fredwso since many websites are using MathJax and so MathML, this allows to get Firefox native MathML.
21:20fredwThere is another add-on to copy the MathML source / TeX source
21:21fredwusing the contextual menu
21:21WG9sa nyone could run the tests and i do run them occaisionally, but I do not run them each day.
21:21fredwand our work-in-progress TeXZilla has been packed into an add-on and accepted by AMO.
21:22fredwSo this means we it can be safely used in other add-ons unlike MathJax or AsciiMath that have &quot;unsafe&quot; rules like eval
21:22WG9stexzila is acdepted. that is great news.
21:23fredwSo hopefully people could rely on this to write new editor-like add-ons
21:24fredwHowever, I the work on TeXZilla is not finished yet. Perhaps raniere you can say more about that?
21:24fredwespecially the integration in MDN
21:24fredw(btw, if you combine the TeXZilla add-on + the MathML copy add-on, you have a way to write MathML equations and paste them into MDN pages)
21:24raniereI was working on a plugin using TeXZilla to MDN
21:25ranierethis week I update it with the lasted version of TeXZilla but found a bug and because of that not release it
21:26fscholzYou mentioned that last time, but I had/have no time to get into that. Sorry
21:26WG9sfredw:so what is the eval issue with Mathjax and AsciiMath is it someithng that can be soved to make sure what is passed to eval is checked to make sure it will nto be a security issue or is it that the reviewers are cluless?
21:26fscholzWhat blocks you on the MDN side?
21:26ranierethis bug is related to the version CKEditor of MDN use (since a start write the plugin from a new version of CKEditor)
21:27fredwWG9s: just because they don&#39;t accept the &quot;eval&quot; call because this is considered unsafe.
21:27fredwand MathJax is too big anyway, one should extract the TeX parser.
21:27ranierefscholz: I hope to fix it this weekend.
21:28fscholzraniere: cool, I hope to have time to have a look. The MDN devs will have a look if you open a pull request :)
21:28raniereYet related with TeXZilla, I&#39;m complete fredw patch to refactor part of the code to improve internal structure
21:28fredwWG9s: it&#39;s not necessarily that there is a security issue. just that they have some rules to accept or reject add-ons. I guess you know that, since you tried to submit add-ons once :-)
21:28WG9smany things are unsae but can be made safe if the thing that calls it validates the input parmeters cannot invoke it in an unsafe manner.
21:29ranierefredw: can you give more details about TeXZilla&#39;s refactor?
21:30WG9sthe issue i had with my add-on hd to do with what it was intended to be. was supposed to be an add-on to be a proof of concept to get code adde to the base because that was aked for int he bug. but the changes required to have it accepted as an add-on were goign to make the code farther from what you would need to do to integrate it into the code.
21:30fredwCurrently, TeXZilla generates the MathML string immediately. Raniere is rewriting that a bit to use more abstract structures and detect duplicate or useless attributes
21:31WG9sadn then after we showed how well the addon worked they decided they didn;t like the feature anyway.
21:32WG9sso the idea wa to get it dded to the product which is not going to happen so I lost interest in geting it accepted on add-ons. so I jsut shot it on my website.
21:32fredwWG9s: I don&#39;t know if you are talking about the &quot;pink&quot; add-on, but I thought you had other add-ons that are available on your website
21:32raniereThe last thing about TeXZilla: I&#39;m planning to create a webapp for Firefox OS as proof of concept for it.
21:32fredwAnyway, I was a bit worry for TeXZilla, since it is parser generated from a grammar by Jison
21:33fredwand it could have been considered obfuscated code
21:33fredwraniere: sounds a good idea
21:34fredwraniere: I think with mobile devices where you typically always have a virtual keyboard, having button to insert special math symbols become something natural
21:35fredwso you could extend the virtual keyboard with summation symbols, square root etc
21:35fredwthat was one of the reason to make TeXZilla accept unicode characters
21:35ranierefredw: thanks for that, this give me a idea for GSOC
21:36fredwso one could e.g. just insert the unicode sigma symbol instead of typing \sigma
21:36ranierebut backing to the agenda, there is the MathML in Wikipedia.
21:37fredwI think Moritz has made some progress on that but he is not present today
21:37fredwyou can check,dashboards/default
21:37WG9sI only have 357 users using the add-on of course. ;-)
21:37fredwfor open changes and those recently merged
21:38fredwWG9s: which add-on
21:39fredwI&#39;m trying to help Moritz by reviewing the patch and trying to make things move forward.
21:39fredw(WG9s: how do you have stats on your add-ons)
21:39WG9snot mathml related thisi is my Ftp_Upload extension. the 357 users is based on number of users access my site to check for updates on it.
21:40fredwWG9s: a ok.
21:40raniereCan we go to Gecko developments?
21:40fredwraniere: so I&#39;m continuing to work on the Open Type MATH
21:41fredwand as always you can check Bill&#39;s builds for the latest version
21:41fredwand try to see how it looks like
21:42WG9sbut i still have no capability to do mac builds
21:42* raniere like the preview
21:42WG9sso only windows linux and android
21:42fredwKarl reviewed the first patch (bug 663740) recently and it should land soon in trunk (hopefully)
21:43fredwJonathan Kew has worked on the bug with ascent/descent
21:43fredwand James fixed the issue with primes
21:43fredwso now the test page should give a relatively good idea of what the changes could look like
21:43fredw(hopefully in Gecko 30?)
21:44raniereI try to work on Bug 947654 - Default fonts for MathML
21:44fredwSo I&#39;ve increased karl&#39;s review queue again with bug 407059
21:45fredwraniere: I&#39;m already working on this one (cf the patches on Bill&#39;s page)
21:45fredwI thought you were working on the automated installation on Linux
21:45WG9sthere still seems to be an issue on windows aith asana math and bug 947650 but the same issue seems to occur with that font using the windows font viewer so hrd to blame it on gecko.
21:46ranierefredw: sorry, yes, the automated installation
21:46fredwWG9s: yes, I think Jonathan is aware of that
21:46fredwbut I told him that it&#39;s ok since the bug does not appear with Cambria Math and this one is installed by default
21:47fredwWG9s: BTW: do you have any plans for Mac?
21:47WG9sbt the fct that the extra vertical spacing shows up with the windows font viewer would seem to indiacte this is either a font issue or a windows issue.
21:47ranierebut since was the first time that I read gecko&#39;s code was very hard do understanding how it works and I fail in my try
21:49fredwWG9s: apparently there are several metrics for ascent/descent and the application has to choose the correct one. Jonathan&#39;s patch was to use the so-called typo metrics for fonts with a MATH table. At least that&#39;s what I understood but karl may correct me if I&#39;m wrong
21:49WG9sfredw:I don;t really have plans for a mac right now. maybe when i retire so this mozilla stuff has more time but that is a year or so away.
21:50fredwraniere: it&#39;s sometimes hard to get involved in Gecko. That&#39;s why I try to propose &quot;easy&quot; bug to newcomers.
21:50* karl doesn&#39;t need to correct
21:51WG9sI am runnign vmware workstaion under windows 7 on one of my build systems. if someone could get me something that could build mac version that soule run under that, i would be willing totry.
21:51fredwWG9s: is it legal to run a mac system on Windows system? :-)
21:52ranierefredw: I was think in write a little about how gecko works to help new developrs like me
21:52raniereanyone know some think like that? ^
21:52fredwI think this is a very good start for newcomers.
21:53fredwMozilla people have done a great job to help that.
21:54ranierefredw: I&#39;m look for something that explain how gecko/firefox render the html (gfx I think)
21:54fredwTo come back to the pad, James Kitchener also has done some work on rowspacing/columnspacing etc and he said this even fixes some issues with RTL mtable
21:54fredwraniere: I think it&#39;s more in layout/ depending on what you mean
21:55fredwI guess James will publish a patch soon, so Bill could include it in his build
21:55fredwWG9s: I also have a patch for tooltip:
21:55fredwbut have not tested it much yet
21:56WG9sspi should maybe try to incoudethat tomorrow?
21:57WG9sor i could respin todays builds with that
21:57WG9sany other bpatches you would like to see to test as long a s I am doing this?
21:58fredwWG9s: I don&#39;t think it&#39;s too important, feel free to choose when you include it. I don&#39;t have any other patches.
21:58fredwhi bruce
21:58WG9sOK I will start builds now with that
21:59ranierebruce: welcome
21:59brucesorry I&#39;m late; got confused by zones....
21:59fredwNo problem. We were just about to talk about LaTeXML
22:00fredwbruce: do you have anything new to share? I wasn&#39;t really sure, so just put the &quot;latex to docx&quot; on the pad
22:00fredwWG9s: thanks
22:01brucelotsa handwaving. still slowly sneaking up on consolodating lots of changes into a proper, overdue, release
22:01fredw(bruce: also, I quickly mentioned the epub conversion and graphics package in LaTeXML last time, but nobody from LaTeXML was here to give details)
22:02bruceYep, there&#39;s a draft implementation generating ePub in there, thanks to Deyan Ginev
22:02fredwSo what is the plan for LaTeXML (release, features included?)
22:02bruceshould be ePub3, btw, with MathML and so on.
22:03brucequite a bit of tikz/pgf graphics stuff works, generating SVG
22:03bruceso can make good html5 now
22:03WG9sOK builds have started with that new patch included
22:03ranierebruce: all great news =)
22:04fredwbruce: BTW, I talked about previously
22:05bruceA couple of Michael Kohlhase&#39;s students from Jacobs (Lukas Kohlhase and Kim Jablonski) have been working on XSLT to generate word/openoffice formats (whether that extends to docx, I don&#39;t quite know yet)
22:05ranierebruce: when I try LaTeXML I had to use imagemagick without HDRI, do you know if it&#39;s possible to use imagemagick WITH HDRI?
22:05fredwbruce: I&#39;m wondering if in the future that would be the opportunity to continue the implementation of the XeTeX package to choose specific fonts
22:06bruceI&#39;m sorry, what is HDRI?
22:07bruceBTW: I&#39;ve managed to extricate the Image::Magick specific stuff so that it is no longer a requirement to install LaTeXML, but without it, you loose the ability to do certain image manipulations; particularly generating images for math (but that&#39;s not necessarily a bad thing, in this crowd!)
22:08ranierebruce: I will test again with the extricate version
22:08fredwbruce: I was thinking about the \usepackage{xunicode} \usepackage{fontspec} mentioned in
22:09fredwwith the new font improvements, it should be possible to easily select arbitrary MATH fonts in Gecko with just a font-family. So that could be something usable in LaTeXML
22:09fscholzHas anyone any questions on MDN docs or the doc status page I shared on the etherpad?
22:10fredwfscholz: perhaps you can talk about it if you wish
22:10fscholzIf not I will step away from the computer in some minutes :)
22:10fscholzYeah, I can.
22:10WG9ssorry if i seem to only be half here i am also listeining in on a 2 day presentation by cisco for my real job.
22:10bruceNot completely clear how it applies; it&#39;ll use whatever image magick you have installed, but it basically is only interested in dealing with images in 2 places: generating an image for math/picture/etc, or dealing with image conversion for \\includegraphics and friends. Perhaps there are special cases of image processing that would be useful in the latter ca
22:10fscholzDont want to interrupt you though
22:11fredwfscholz: So while we are on MDN, do you have any plan for a GSOC project?
22:12bruceAs far as xetex, I got to find some time to play with xetex itself soon, and see what ways that latexml isn&#39;t doing the right thing; often there are special versions of packages for xetex that wouldn&#39;t be known to latexml
22:12fscholzfredw: No I think we havent yet. teoli organized it last year
22:13bruceotherwise, since the underlying perl does unicode, but latexml inserts a layer that pretends it&#39;s 1980, when you then want to emulate xetex, not sure whether we want to add or subtract a layer of abstraction!
22:14fscholzSo, for documentation status, we are looking into creating sections on MDN and track their status. MathML is such a section.
22:14fscholz tracks everyting under
22:14fscholzIn the table you can see some metrics / &quot;health indicators&quot;
22:15fredw(ok, I&#39;m asking that because the brainstorming is open: So if anyone wants to propose an idea and mentor a student for a MathML project, feel free to contribute)
22:15fscholzThere are two columns for tag related tasks, two for review tasks, open documentation request bugs and dev-doc-needed
22:16fscholzIn the future we will add more health indicators as we implement them
22:16fredwfscholz: does that track only the Web/MathML section
22:16fredwnot the Project/ section?
22:16fscholzAt the very bottom (and that is a call to action for you) there is a todo/ wishlist
22:16fscholzPlease add ideas for docs we want to have there
22:17fscholzfredw: yes, it only tracks that part currently
22:18fscholzWe need to look into that. Most projects have their project docs on and web developer / end user documentation on MDN
22:18fscholzBut it is a bit grey, there is no right and wrong imho
22:18fredwfscholz: what is the difference between &quot;Documentation requests&quot; and &quot;Dev-Doc-Needed bugs&quot;?
22:19fscholzThere is &quot;Developer Documentation&quot; product on Bugzilla and that has a &quot;MathML&quot; component
22:19fredwalso, I&#39;m not sure who is in charge for the technical/editorial review for MathML
22:20fscholzThere are no bugs filed currently
22:20fscholzfredw: It&#39;s community driven, we usually dont have people in charge
22:20fredwfscholz: OK, so that&#39;s when we want to open a bug to request a more complete doc than just release info on bug fixes and new features
22:21fscholzBut we want to make these tasks more visible so that people are aware of it
22:21fredwfscholz: so how do we &quot;review&quot; the Deriving the Quadratic Formula for example?
22:21fscholzFor example, yes.
22:22fscholzTechnical review means the content is technically ok. That is the MathML markup makes sense and the formulas as well
22:22fscholzeditorial review would be wording and grammar
22:23fredwmmh, <mn>- c</mn> does not look good
22:23fredwfscholz: ok, thanks for the info
22:23WG9sfredw: on bug 969906 fr tooltips do you have a testcase?
22:23fscholzYeah, someone who understands mathml technically like you do, can review it
22:23fredwI just logged in and see a &quot;submit review&quot;
22:24fredwI propose to quickly finish with the other subjects on the pad
22:25ranierefredw: do you mean gecko development?
22:25fscholzFeel free to ask me here or email if you have questions on that.
22:26fredw&quot;miscellaneous&quot;, I think Gecko dev was addressed
22:26fredwfscholz: ok thank you
22:26ranierefscholz: thanks
22:26fredwfscholz: I&#39;ll try to give it a try when I have time
22:26raniereMozilla was working in a version of Firefox OS for tablets and look like they going to start tests
22:27fscholzfwiw, a more populated doc status page looks like this
22:27ranieremaybe some one can try get one of the developers device to test MathML support
22:28fredwraniere: I believe will be important
22:28ranierefredw can use that device to test his ebook
22:29fredwonce bug 947654 is fixed, we should try to get math fonts installed by default
22:29fredwraniere: did you find an e-reader for FirefoxOS that supports Javascript?
22:30ranierefredw: no =(
22:30fredw(BTW, one of the important MathML feature for mobile device will be linebreaking
22:31fredwraniere: OK. I wonder whether they have an e-reader installed by default.
22:31ranierefredw: there is no official epub reader yet
22:32fredwIMHO, they should do like Apple&#39;s ireader which reads PDF and epub and is installed by default.
22:32raniereit has pdf.js installed by default
22:33ranieremaybe it can be change to use epub backend support?
22:34fredwraniere: I think a single application should be installed by default to read PDF ebook or EPUB ebooks like Apple does
22:34fredwraniere: so your plan was to get a developer device?
22:35ranierefredw: at least try. they will be send some devices for developers and I believe that you had great chances to get one
22:36fredwraniere: I have not checked in details the procedure. I just saw a post on planet mozilla
22:36ranierethere is just the post on planet until now.
22:36fredw(BTW: raniere and fscholz &#39;s blogs are now on planet mozilla)
22:37fredwOK, I&#39;m getting sleepy
22:37raniereThe last topic was GSOC (Google Summer of Code)
22:38fredwI quickly mentioned it
22:38ranierefredw: implementing the math keybord can is a suitable project?
22:38ranieres/can /
22:38fredwraniere: I guess so. But I don&#39;t know about FirefoxOS at all
22:39ranierefredw: no problem. I have some friends here in Brazil that I can ask.
22:39fredwraniere: so do want to apply as a mentor?
22:40ranierefredw: can I apply as a student?
22:40fredwanother idea for a GSOC project / Addon was to enhance thunderbird with math features:
22:40fredwraniere: if you are a student, yes
22:40ranierefredw: yes, I&#39;m a student. =)
22:40fredwbut if you don&#39;t have a mentor for the project, you are unlikely to be accepted
22:41fredwso perhaps you should contact some FirefoxOS people first
22:41raniereI think that I can find a mentor for Firefox OS keyboard.
22:42ranierefredw: I will do that
22:42fredwand that will be integrating TeXZilla too I guess
22:42ranierefredw: sure!
22:42fredwor some kind of MathML apps
22:43raniereAnyone have one last topic?
22:43fredwI think that&#39;s all for me
22:43raniereWG9s: one last topic?
22:44fredwbruce, karl, teoli?
22:46ranierefredw: bruce: karl: teoli: WG9s: hope to see all next month.
22:46fredwok, thanks for leading the discussion
22:46bruceOK, talk to you then; bye!
22:47raniereI&#39;m closing this meeting. And will write the blog post about it later.
22:47fredwok, great. Normally it should show up on planet mozilla this time
22:47fredwgood bye
22:48ranierefredw: good bye
22:53WG9sfredw: linux64 -bit build with patch for bug 969906 is on my site others should be there within an hour or so.
23:29WG9sonly waiting for linux 32-bit now.
14 Feb 2014
No messages
Last message: 167 days and 16 hours ago