Perl 6 - the future is here, just unevenly distributed

IRC log for #webwork, 2014-05-01

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

All times shown according to UTC.

Time Nick Message
00:20 rbeezer joined #webwork
00:45 goehle joined #webwork
01:34 mgage joined #webwork
02:23 mgage joined #webwork
12:22 mgage joined #webwork
12:36 goehle joined #webwork
13:18 mgage joined #webwork
13:21 goehle hey mgage
13:24 mgage goehle:  hi
13:37 mgage goehle — hi — I’m around until about 9:50am — then I have a thesis committee meeting
14:53 rbeezer joined #webwork
15:46 aubreyja joined #webwork
15:46 aubreyja joined #webwork
16:02 mgage joined #webwork
16:06 goehle hi mgage
16:14 mgage hi
16:14 mgage goehle: hi
16:16 goehle hey
16:16 goehle finally on at the same time
16:16 goehle so I've been considering pulling back my initial plans even farther and abandoning the unique id upgrade plan
16:16 mgage yeah — I’m running around like crazy these last two days.
16:16 goehle I would still need to add a unique id to the problem table to get my stuff tow ork
16:16 goehle but it would be on top of the existing things and wouldn't change how problems work now
16:17 mgage ok
16:17 goehle I wouldn't add unique ids for any of the other tables and they wouldn't be integral
16:17 mgage ok — seems like we could add those later if it works out well with the problem table
16:17 goehle well
16:17 mgage is the problem table alone or the problem/user table
16:17 goehle see I"m not making it an integral part of the problem table
16:18 goehle it would be a field for the problem table alone
16:18 goehle and it wouldn't be a primary key or any of the other stuff I did in the first pass at the jitar branch
16:18 goehle it would be closer to the psvn field
16:18 goehle something extra that can be used
16:19 goehle what got me thinking about this is that changing the id's like I did breaks pretty much almost everything
16:19 goehle webworksaop
16:19 goehle the ajax interface
16:19 goehle all of peters stuff
16:19 mgage ok  — just checking again — this is for the problem itself — and places something like the role of path-to-file for identifying it?
16:19 goehle no
16:19 goehle this would be for globalproblem
16:19 goehle but would be independent of the source path or anything like that
16:20 goehle its because problem sets are trees in jitar
16:20 goehle and I need something more robust than the problem number (which is currently what problem_id is) in order to identify things
16:20 mgage right but it to first approximation the unique_id and the path to source would identify the same problem
16:20 goehle not if the same path to source was used in multiple problems
16:20 mgage right — the problem number within a set has never been a very good id
16:21 goehle which is certainly something people are thinking about
16:21 goehle (or more likely if they are both the same .def file for a problem pool)
16:21 mgage ok — so the same source might be used in several problems with different uniqu_ids
16:21 goehle right
16:22 mgage and the unique_ids are kind of unique to a site, or are we trying for something that is unique across sites.
16:22 goehle unique to a course even
16:22 mgage ok
16:22 goehle this would be pretty much completely abandoning the old stuff
16:23 goehle just the bare minimum of what I need in order to make JITAR work
16:23 goehle in particular I wouldn't be designing this unique id with any other applications in mind
16:24 mgage ok.  I haven’t had time to look over the earlier plans in detail — I’ve no problem with doing what is necessary for the moment, but at some point I’d like to look at the earlier plans and figure out what could be accomplished with minimum breakage.
16:27 goehle well
16:27 goehle if its a matter of fixing the way we use id's
16:27 goehle then that touches on pretty much everything
16:27 goehle a lot of it would be easy to "fix"
16:28 goehle but almost everything would need some tweaking
16:28 goehle so my dilemma is this
16:29 goehle is it worth doing the id fix (and fixing the resulting mess) if that is *all* we are doing
16:29 goehle I think its a worthy thing
16:29 mgage that sounds right.  it isn’t immediately needed but I”d still like to look to see if it’s possible or desirable to do.  I’ll need to block of time to look at it which I don’t have right at the moment
16:29 goehle and I did want to use this project as an opportunity to really improve some stuff
16:30 mgage as a first step in your project I think it’s ok to do the minimum and get things working.
16:30 goehle well, I'm making choices that the project will have to stick with
16:31 goehle its not like the stuff I do now will scaffold into later better changes
16:31 mgage if I see a block of time and some people to work on it we can try to do a concentrated change and fix the mess.   it would be a separate branch all on it’s own with just one focus.
16:31 mgage brb
16:31 mgage student  — I’ll be back in a bit.
17:06 mgage joined #webwork
17:07 mgage back
17:08 goehle hey
17:08 mgage I guess I don’t have a good feeling for whcih choices you are making are locking us into a data model that is more restrictive than what we already have.
17:08 goehle its not that its more restrictive than what we have
17:09 goehle its just adding another layer on to what would need to be redone if we start redoing things
17:10 mgage ok .  so my take is that if we have time to both create the JIT module and revamp the old one then that would be a good idea — if we don’t have that time and manpower then we have to get the JIT module working with what we have and kick the bigger changes into the future.
17:10 goehle right
17:10 goehle so I'm trying to figure out which of those two I should do
17:11 goehle along with that
17:11 goehle is it worth it to redo the id stuff
17:11 goehle just by itself
17:11 mgage there is the danger of doing that forever — we’ve already lived with it for longer than I planned, but if the alternative is to make everything not work for 6 months then this is not the right time to do that
17:11 goehle i.e. if we are going to be messing with everyting, should we be more ambitious
17:11 goehle well it won't affect webwork2 or even webwork3
17:12 goehle and the main thing is I could probably get the JITAR stuff working sooner
17:12 goehle so they could run JITAR on a custom server
17:12 goehle but not use things like moodle integration or any other features that haven't been upgraded yet
17:12 goehle (they have to use math4, classlist editor 2, etc...)
17:12 mgage that’s a delicate question and I have to study more to give an answer.  sometimes it’s worthwhile to change one thing, get information, then change the next — ven though it seems like it is more work (and probably is) it’s a more stable process
17:14 mgage I lean toward getting the JITAR thing running sooner — particularly if it is a lot sooner.  if you can do something like unique_ids and only delay development for a few days then that’s probably worth it
17:15 mgage because you are already looking at the code and are famililar with it — if that change is going to require lots of other changes as well then you have to make that there is enough man-hours from someone to track down all the little inconsistencies
17:15 mgage otherwise you’ll get distracted from the main project.
17:16 goehle ok
17:16 goehle here is a fun thing that I need for JITAR, btw
17:16 goehle $o=1;{grep!($o%$_),2..$o++or print$o;redo}
17:16 goehle a 42 character perl line for generating the nth prime
17:17 mgage cool
17:18 goehle ok
17:20 goehle well that seems to decide it then
17:20 goehle I'll abandon the id refactoring for now
17:21 mgage ok.  I’ll have more time to concentrate on this in a bout a week.  I’m in the midst of  exams and senior checks right at the moment.
17:21 goehle sure
17:21 goehle our finals are next week
17:22 goehle lots of activity on the github lately...
17:22 goehle I guess thats good :)
17:23 mgage yes — and thanks very much to you and to Peter for taking care of all of that — I really have not had a chance to look at it, except to observe that it is taking place.  It’s been a surprisingly busy semester for me. (teaching point set topology takes a lot of work :-) )
17:24 goehle true
17:24 goehle ok, well I'll let you get back to it
17:24 goehle ttyl
17:24 mgage ttyl

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