Perl 6 - the future is here, just unevenly distributed

IRC log for #webwork, 2014-09-22

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

All times shown according to UTC.

Time Nick Message
02:35 mgage joined #webwork
12:46 mgage joined #webwork
13:03 goehle joined #webwork
14:18 rbeezer joined #webwork
14:20 shdw_work joined #webwork
14:47 rbeezer_ joined #webwork
15:24 aubreyja joined #webwork
15:24 aubreyja joined #webwork
15:34 mgage joined #webwork
18:08 goehle hey mgage
18:08 goehle as you can see I got up a patch for the encoding stuff, but its for 2.9
18:08 mgage hi geoff
18:09 goehle I backported a reduced version to 2.8+, but I had to do it without any of the formal pull request stuff because github doesn't support "pulling to tags"  (that doesn't actually make sense)
18:09 mgage I saw that I haven’t pulled it yet
18:09 goehle it definitely needs to be tested, but I iterated a couple of times and its not the hugest change
18:11 mgage right — let’s check with Davide who has actually done this but I think one can download the master branch to your local git, insert your fix in 2.8 and move the 2.*+ tag and then push this change back up to github.   — not something we want to do often — but in this case it might be a good idea.
18:11 goehle I did do that
18:11 goehle I just don't have a pull request for it
18:12 goehle I also did a much smaller set of changes, since they were not going to be externally tested
18:12 goehle I describe more in the PS to the pull request for the main change
18:12 goehle so PG-2.8.1+ is updated.  If someone using 2.8 wants to upgrade they should be able to by merging that tag.
18:12 mgage ok — I think that is the best we can do.  — I think it’s good enough
18:13 mgage I’ll have a chance to try that. — the main server here is running 2.8
18:13 goehle ok
18:13 goehle that just adds more character escaping to the existing escaping in PGbasicmacros.pl
18:13 goehle so its not as complete (it only does stuff that goes through basicmacros)
18:13 goehle but you won't even have to restart your server
18:14 mgage should be enough.  As far as I know no-one has noticed this exploit in over a year.
18:14 goehle lets keep hoping :)
18:15 goehle oh I was going to mention the other day
18:15 goehle remember I had this "clever" idea to store a jitar problem number (like 2.1.3) using prime powers as 2^2*3^1*5^3
18:16 goehle I put clever in quotes because of course those numbers grow exponentially and you can only fit about 32 of them into a set before you get overflow errors
18:16 mgage I remember
18:17 goehle It turns out the reason why cs people dont use that is because they do a much better thing with bit masking
18:17 goehle So now I use logical shifts, ors, ands, and bitmasks to store the address in a 32 bit integer as
18:17 goehle SAAAAAAABBBBBBCCCCCCDDDDEEEEFFFF
18:18 goehle where S is the sign (ignored) A is the first address integer, B is the second, and so on
18:18 goehle its fancy
18:18 mgage not bad.  perl has lots of bit level commands — I’ve seldom used them
18:18 goehle basically its a custom version of how ip addresses work
18:18 goehle neither have I.  I think this is the first time I've really needed to go to bit level to do something
18:19 mgage we used these ideas more in the very early days when using GDBM storage we only had 1028 bytes per record. :-)
18:19 goehle right
18:19 goehle and even now the more appropriate thing would be to just store it as a string
18:19 goehle but I didn't want to change the column type of problem_id
18:20 mgage even then I didn’t use bitcodes very much — but key names were severly abbreviated.
18:56 mgage joined #webwork
20:54 soumya joined #webwork
22:17 mgage joined #webwork
23:48 Brando753 joined #webwork

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