Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-07-04

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

All times shown according to UTC.

Time Nick Message
00:28 sivoais here's a research implementation of APL that compiles to an intermediate representation
00:28 sivoais <https://github.com/melsman/apltail>
00:28 sivoais there's a paper that takes that intermediate representation and targets it to GPUs
00:31 sivoais <http://conf.researchr.org/track/pldi2015/ARRAY-2015-papers#event-overview>
06:11 opkick [pdl] zmughal pushed 4 new commits to sf390: http://git.io/vqtuh
06:11 opkick pdl/sf390 d2cb5f9 Zakariyya Mughal: test scalar comparison across several values on either side of bad value
06:11 opkick pdl/sf390 ce91147 Zakariyya Mughal: clean up BadCode string for `biop()` codegen
06:11 opkick pdl/sf390 0737e60 Zakariyya Mughal: test fix: the result of the comparison ops should set the badflag
06:29 sivoais why does ... gah...
06:30 sivoais > use PDL; $a = pdl(2); $b = pdl(5); $b->badvalue
06:30 sivoais -1.79769313486232e+308
06:30 sivoais > $a->badvalue(5)
06:30 sivoais 5
06:30 sivoais > $b->badvalue
06:30 sivoais 5
06:31 sivoais why does setting one badvalue set the other? O_O
06:44 sivoais I guess it is through propagate_badvalue
06:45 sivoais I haven't gone into this (what seems) undocumented portion of PDL
07:11 sivoais ah, nvm... I understand
07:11 sivoais I had thought I had PerPdl badvalues compiled in
07:17 sivoais but funnily enough, I was coding thinking I had it compiled in, so I wrote code for that case...
07:17 sivoais and the Travis-CI build indicates that I fixed the bug for that <https://travis-ci.org/PDLPorters/pdl/builds/69517571>
07:41 opkick [pdl] zmughal comment on issue #130: Adding to the devops task list for the split as mentioned in #119. http://git.io/vqtiB
07:43 opkick [devops] zmughal pushed 1 new commit to master: http://git.io/vqti9
07:43 opkick devops/master bd45935 Zakariyya Mughal: add note about Appveyor to the task list
19:19 chm joined #pdl
19:35 mohawk chm, evening!
19:36 chm hi mohawk
19:37 chm I'm trying to catch up with work by sivoais
19:38 chm I saw the blog from david mertens and I hope that means he has a job...
19:38 chm :-)
19:40 chm sivoais: regarding PDL3 thoughts, rather than updating PDL core and the PDL::PP my thought was build the computation foundation first
19:41 chm If you look at the places where PDL-2.x has difficulties in implementation or features, alot comes from the limited data type suport
19:42 chm however, I would rather replace "lowest-level possible" with most fundamental level.
19:42 chm At that level, PDL-2.x provides data representation and support for efficient computations with the same
19:43 chm Ideally, the basic PDL types should include elemental types, structure/object types, ...
19:44 chm Similarly, the [auto]looping/threading operations range from strided, to indexed, to arbitrary...
19:45 chm Recognizing that fact and enabling those concepts and possible extensions in the PDL core would greatly help PDL3
19:48 sivoais right,  chm. agree 100%. I have some written and mental notes which I will post later,  but I can't this month. lots of work to do on my end
19:49 sivoais but let me know if the changes i made help with sf390
19:49 sivoais :-)
19:54 sivoais oh, and I know that one of my tests is misguided. The stats test that I wrote only passes on per pdl bad values
19:55 sivoais but it shows a pitfall when setting the bad value directly
21:07 sivoais joined #pdl
22:23 mohawk sivoais, if we establish that sf390 has been around for ages (which i think is what you said), what do you think of pressing for us to continue with the big split?
22:43 sivoais mohawk: chm really wanted to get this squared away first
22:43 sivoais it's close to being fixed, I just need some feedback on the semantics of badvalues
22:43 sivoais which I asked for in my e-mail
22:45 sivoais <http://permalink.gmane.org/gmane.comp.lang.perl.pdl.devel/6579>
22:46 * sivoais is going to be afk
23:04 mohawk sivoais, did i understand you right that you're working on a 390 fix, but you don't have time to work on it this month?
23:10 sivoais oh, I am going to fix it, I don't have time to write up my thoughts on pdl3
23:12 mohawk glad to hear it
23:13 mohawk let us focus our resources

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