Perl 6 - the future is here, just unevenly distributed

IRC log for #webwork, 2013-05-29

| Channels | #webwork index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:40 mgage_ joined #webwork
13:32 mgage_ joined #webwork
14:46 mgage_ joined #webwork
15:49 mgage_ joined #webwork
17:34 holliday joined #webwork
17:38 goehle joined #webwork
17:39 goehle hey holliday
17:41 holliday hey geoff
17:41 goehle hey
17:41 goehle this is the irc
17:41 holliday I see
18:41 goehle_ joined #webwork
18:41 goehle_ hey mgage
19:00 ionparticle joined #webwork
19:17 mgage_ joined #webwork
19:40 mgage_ joined #webwork
19:46 goehle hey mgage_
19:46 mgage_ hi
19:47 goehle was just talking to Mark, the CS guy who will be coming to AnnArbor
19:47 goehle and I have a whole series of questions for you, if you have the time :)
19:47 mgage_ sure -- by the way I entered your blog but I just noticed that I got the address wrong -- I've fixed it but it may take a while before the page updates
19:48 goehle no worries
19:48 goehle first, do you know why permissionLevel and Password are separate tables (and not part of user)
19:49 goehle ?
19:52 mgage_ mostly historical -- the very first database GDBM could only hold 1K of characters in a record.
19:53 goehle do you know what the depths table is for?
19:53 mgage_ that said if we go with SQL tables it seems desirable to have tall, narrow tables in general and then get "full" records using left joins.
19:54 goehle I was talking to Mark about that
19:54 goehle to make that efficient we would need to seperate out
19:54 goehle user_id
19:54 goehle from user_name
19:54 mgage_ yes -- it is a global table added much later to supply information on how high to offset images when typesetting mathematics.
19:54 goehle searching on blobs is slow I guess
19:55 mgage_ Yes.  I think also that we want to make all the tables global, not related to a specific course -- this makes updating somewhat easier it also makes it easier to add features with new modules.  It makes archiving courses more complicated (you can't just save individual tables) and it might or might not have performance issues.
19:56 mgage_ I don't know about the last but I know that at least about 6 or 7 years ago programs like moodle adopted the tall thin model claiming greater efficiency
19:57 mgage_ not sure what the current opinion is
19:57 goehle some of the tables (like past_answer) will get really big if you do that
19:57 goehle and we would really need to fix the id/name issue
19:57 goehle which is a lot of work but coudl definitely be worth it
19:57 mgage_ true but that is what sql tables are supposed to be good at.  it would require appropriate indexing
19:58 mgage_ resolving the user id and user name issue would be part of it
19:58 goehle this sounds like a good discussion for AnnArbor :)
19:58 goehle as a side project I was thinking about removing tabber from everything and just putting in the bootstrab tabs
19:59 mgage_ in fact *every* record in every table would have a timestamp and a unique id whether it needs it or not. :-)
19:59 goehle any issues with that
19:59 mgage_ no the twitter bootstrap stuff works fine
20:00 goehle especially since I already added some basic bootstrap stuff to math3 for backwards compatibility of the modal features
20:00 mgage_ the unique id would be used for indexing in the tables and nothing else
20:01 mgage_ right -- you might have to beef up math3 a little to keep it usable.  Math2 is pretty much the same as math3 but it uses serif fonts instead of non-serif fonts
20:02 mgage_ ok -- your feed works now
20:02 mgage_ to the planet aggregator
20:02 goehle great
20:03 mgage_ by the way -- that part of your work which is in release/2.7 is working on hosted2 now if you want to stress test it.
20:03 goehle nice
20:03 goehle I'll also pull it to my test server
20:03 goehle math4 is looking sharp
20:04 goehle but it has issues with olderversions of ie
20:04 mgage_ the latest achievement items get to wait for the next release.
20:04 goehle yeah
20:04 goehle they should
20:04 mgage_ how old a version of i.e.?
20:04 holliday joined #webwork
20:05 goehle its fine on 9
20:05 goehle on 8 its serviceable but looks like of ugly
20:05 goehle and there are real problems on 7
20:06 mgage_ ok.  I wouldn't worry about it too much but that is a good  reason to keep math2 and math3 around for a while.
20:06 goehle yeah, although users don't get to pick the theme
20:06 goehle only instructors
20:06 mgage_ a good short term project would be to hack the ability to choose a theme in to the section where you choose which method to use to display mathematics.
20:07 mgage_ eventually that info should be stored in a "session preferences
20:07 goehle yeah
20:07 goehle I was also thinking it might be good to take a stab at getting webwork pages to validate
20:07 goehle although that may be a long row to hoe
20:07 mgage_ table (or perhaps a user preferences table) but it could also be stored on the page in the same way the choice of displayMode is stored. (this is not a good idea long term but it would work for now.)
20:08 mgage_ yes.  I'm curious how far off we are.
20:10 mgage_ there was also talk about using something besides CGI.pm for HTML entity creation (we don't use CGI.pm for much more than that).  I'm not sure what the alternatives are Template::Toolkit and Template::HTML  perhaps?  We should be emitting less HTML from the perl pages in general but I think we still want to emit some HTML tags and it would be best if that was generated in a central location so that it can be updated.
20:14 goehle I saw jasons email
20:15 goehle whats wrong with CGI?
20:15 goehle I didn't read the perl guru blog thing
20:15 mgage_ yes -- that's what I was referring to.  CGI does a lot of other things besides generate HTML.  And we're not using that.
20:16 goehle if we aren't using that is it such a big deal?  Getting rid of CGI would be  a chore, but not hard
20:16 mgage_ There are better choices for doing the other stuff.  And pages are written these days with less detailed tags and more CSSl and javaScript gloss.  -- that's feeding a little bit of the discussion -- and the push to get rid of it might be a little over hyped.
20:17 goehle It could be a good opportunity though to try and break the html generation away from the content generation
20:17 mgage_ but a light weight way to generate HTML might still be attractive for us.
20:17 goehle righ
20:17 goehle I imagine one big project would be to try to take the existing code
20:17 goehle move all of the heavy duty html out into template/js land
20:18 goehle and then whats left will use light weight code to generate vanilla html tags
20:18 mgage_ some of the HTML generation should be with the content generation -- it seems to me -- particularly since we are doing dynamic generation so we have a lot of control of how that stuff is generated.
20:18 goehle yeah
20:18 goehle I mean, when you output a table the tags should be there
20:18 goehle and there should be div's with id's and classes on them
20:18 mgage_ so input blanks should have HTML <input type="text"> tags
20:18 goehle righ
20:19 goehle I just tried to validate a few pages
20:19 goehle its not as bad as you might thing
20:19 goehle only a few errors and they are pretty minor
20:19 mgage_ at the moment we''re not generating a lot of extraneous HTML (at leasts from the problems)
20:20 mgage_ that's kind of what I expected -- and for the problem pages themselves a few fixes in the code in strategic macros will fix a lot of apparently separate errors.
20:20 mgage_ I think it's a pretty doable project.
20:21 mgage_ The model at the moment is that when we generate a problem we know what display mode it's aimed at -- TeX, or HTML with MathJax or with images (dvipng) so we can customize the display.
20:21 goehle yeah
20:21 goehle so fixing some macros would help with problems
20:22 goehle and errors in the other html just need to be tracked down in the perl
20:22 mgage_ Creating a static version of the problem and then using javaScript to convert it to TeX or MathJax or dvipng has seemed to me to be overly complicated -- it has been suggested (some years ago in fact) , but I haven't been convinced yet that it offers advantages
20:22 goehle hmm
20:23 goehle I mean, you kind of already do that, the static version is the text in the pg file
20:24 mgage_ yes -- that's right. -- the suggestion was to create some xml version of that page and then process it get the output.
20:24 goehle it seems wierd to take existing TeX, translate it into something else, and then get TeX back using js
20:24 goehle ah
20:24 goehle well for a xml was where everyone wanted to go
20:24 goehle even with tex/word docs/spreadsheets/whatever
20:25 mgage_ yes -- that was part of the motivation.  I just didn't see the advantage at the time -- tex is still the most efficient input method for mathematics as far as I'm concerned. How you store it is a separate problem.
20:26 goehle I guess the only advantage would be if there was ever another display method (besides tex/html) then having the xml aroudn would make using that display method really easy
20:26 mgage_ I know publishers moved toward xml schemes for a while, I'm not sure what the state of the art opinion is at the moment.
20:26 mgage_ perhaps -- although adding a new display mode and adding code to 20 or so key macros in PGbascimacros.pl would accomplish the same thing.
20:27 goehle yeah, and I can't even think of what that display mode would be.  Tex/html covers a pretty wide range
20:27 mgage_ audio output -- currently that is being done through MathJax -- but potentially there could be another route.
20:27 goehle ah
20:28 goehle yeah
20:28 mgage_ or generate code for a braille output machine -- again that can be done through MathJax but it could be done directly
20:28 goehle well, I have to head out.  Just so you are aware, after reading up on it bootstrap seems to be pretty broken in general in ie7.
20:29 mgage_ the disadvantage is that modifying PGbasicmacros you need to know something about how that code is written and you probably can't take advantage of a general purpose software.
20:29 goehle yeah.  Alhtough outputing xml that standard enough for off the shelf software to understand doesn't seem trivial
20:30 mgage_ on the other hand to take advantage of the general purpose software you need a well established format that the general purpose software is likely to use for input -- at the moment something like  MathML comes the closest but it's not that close.
20:30 goehle :D
20:31 mgage_ anyway -- I think that particular idea's time has not come yet.
20:32 mgage_ If we get some kind of universal problem format that looks like it will really take off we can modify the basic macros to emit that and let the rest of it be handled by the general purpose stuff.
20:32 goehle I'm inclined to agree.  I think the database stuff is really good to think about.  And I think at some point integrating some web framework ideas could be good
20:33 mgage_ A reasonable way to test this out for now would be to look at the requirements of the moodle question module -- we had a hacked version that would take standard PG output, scrape it a little, and deliver it to moodle for the question type.
20:33 mgage_ If we looked at the requirements for that for moodle 2.x we could see what would be reasonable changes in PG that would be produce output that was suitable for using both in moodle 2.x through it's question module and also through our current webwork2 CMS
20:34 goehle that would be nice
20:34 mgage_ the web framework ideas are likely to arrive with Peter Staab's implementation of instructor tools
20:34 mgage_ backbone.js. and bootstrap.css
20:35 mgage_ that's in development now -- it still has some performance issues.
20:35 goehle I'll have to take another look at those and see where they are
20:35 mgage_ and a lot of ui refinement
20:35 ionparticle hi
20:35 goehle hey
20:35 mgage_ hi John
20:36 mgage_ ionparticle:  I've been trying to get hold of Djun for a couple of days -- do you know if he's around?
20:36 ionparticle just want to put in that I was targeting IE8+ with math4
20:36 goehle yeah
20:37 ionparticle no, sorry
20:37 goehle bootstrap in general is broken with ie7
20:37 goehle It doesn't bother me
20:37 goehle but we should keep it in mind
20:37 mgage_ cool -- in that case the stuff on hosted2 should work well on ie8
20:38 goehle in 2013 IE7 was only 1% of all browser usage
20:38 mgage_ I've added ionparticle's additions to release/2.7 as well and they should all be active on the hosted2 maa1xx courses.  let me know if you see any glitches.
20:39 mgage_ and IE6 was 50%? :-)
20:39 goehle ?
20:39 goehle O.o
20:39 goehle http://www.w3schools.com/browsers/browsers_stats.asp
20:39 goehle I'll poke around during my flight out.
20:40 goehle See you guys in a few
20:40 goehle ttyl
20:40 goehle left #webwork
20:40 mgage_ glad to see IE6 is bowing out finally.  :-)
20:43 ionparticle haha
20:43 ionparticle yeah
20:44 ionparticle btw, Mike, Djun is no longer at UBC
20:47 mgage_ I knew his contract was up -- I hadn't heard what he was up to next.  Also the plan was to have the Vancouver code camp June 27 - 30 at UBC stats department.  I haven't heard an update recently about that and the hotel rooms and want to offer any help I can -- if it's needed.
20:47 ionparticle he's working at a startup now
20:47 mgage_ that means busy :-)
20:47 ionparticle yeah, lol
20:48 mgage_ but I do need to talk to him and offer to organize help if he needs it for the code camp
20:48 ionparticle could try to tweet him
20:48 ionparticle https://twitter.com/djun_kim
20:49 mgage_ kk -- thanks -- I'll start expanding my social media connections.
20:50 mgage_ I know Nora Franzova at langara College -- she was a grad student here at Rochester.
20:51 mgage_ Had you heard about the Vancouver code camp?  I know we were hoping you and Pan Luo could attend -- perhaps there were others also?
20:51 ionparticle sorry, I'm not in the loop with the code camp stuff
20:51 ionparticle I was just going to show up
20:51 ionparticle lol
20:52 ionparticle I'll ask Pan
20:52 mgage_ I'll put both of you on the google docs distribution list so you can see things as they develop.
20:53 ionparticle there's no one else doing webwork dev besides us and math/stats here
20:54 ionparticle are there any plans for overhauling gateway quizzes?
20:55 mgage_ there may be some preliminary discussion of this at Ann Arbor this weekend -- Gavin Larose is the chief author of those.
20:57 mgage_ I shared the Ann Arbor doc page with you as well.  What did you have in mind for gateway changes?
20:59 mgage_ if you have detailed ideas you could put them up on that doc and we'll see what we can get to.  we could continue to work on it at vancouver.  one of the things that would make gateway quizzes easier would be some changes in the database and the conception of what a "set" is.  It was originally a homework set and then gateway quizzes were grafted on, but we should be able to come up with a more general concept that works for both witho
20:59 mgage_ quite so much hacking.
21:00 mgage_ I don't have  an email address for Pan -- if you can send me one I can add him to the doc list as well.
21:04 mgage_ database discussions are planned at Ann Arbor, but i'm sure they won't get concluded there so there will be further discussion at Vancouver.
21:05 holliday left #webwork
21:07 ionparticle looks like I can share it
21:07 ionparticle so I've added him
21:07 mgage_ ok -- thanks.
23:05 mgage_ joined #webwork

| Channels | #webwork index | Today | | Search | Google Search | Plain-Text | summary