Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-03-22

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

All times shown according to UTC.

Time Nick Message
01:06 mohawk (i'm having a go because the coveralls, https://coveralls.io/builds/2166166 now shows lib/PDL/Graphics/PGPLOT/Window.pm as the largest holder of "missed" (untested) lines)
01:51 opkick [pdl] mohawk2 created test-pgplot (+3 new commits): http://git.io/hyYz
01:51 opkick pdl/test-pgplot f3892ef Ed J: Only build PDL::Graphics::PGPLOT if PGPLOT installed
01:51 opkick pdl/test-pgplot 3e17b62 Ed J: Tidy t/pgplot.t, allow window to show if PDL_INT set
01:51 opkick pdl/test-pgplot 4939e88 Ed J: Add t/pgplot.t test with 2 windows
01:55 travis-ci PDLPorters/pdl#152 (master - f879fb9 : Ed J): The build passed.
01:55 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/8467166ef779...f879fb9f4ffc
01:55 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/55329697
02:22 mohawk so i've got a small range of commits that caused the win32 crashes, from derek's clang warnings fixup
02:25 mohawk bisecting further...
02:54 mohawk the guilty commit is b2a7008
02:54 mohawk https://github.com/PDLPorters/pdl/commit/​b2a700884cc5d9729d3e1f86388de95563f5da97
02:54 mohawk which just adds  -Wno-literal-conversion to the CFLAGS
02:55 mohawk which is causing gcc to burst into flames
02:55 mohawk (to the CCFLAGS in fact)
03:09 sivoais mohawk: ah, nice find
03:12 sivoais mohawk: will we need to check the compiler we're using?
03:25 mohawk yes
03:25 mohawk i'm currently getting a fail on t/pthread_auto.t
03:26 mohawk which then passes 2nd time
03:26 mohawk right
03:26 mohawk i'm pushing this change now
03:28 sivoais :-)
03:29 mohawk soon as i've run the test suite one more time...
03:29 mohawk why DO we run prove -v?
03:29 mohawk it produces a vast amount of stuff we don't care about, obscuring any failures
03:30 mohawk well that test is fairly ropy
03:30 mohawk one time fails, once passes, once crashes the interpreter
03:32 opkick [pdl] mohawk2 pushed 1 new commit to master: http://git.io/hy4l
03:32 opkick pdl/master 42f08b4 Ed J: Make -Wno-literal-conversion clang only - crashes gcc 4.8.2 Win64
03:32 mohawk all done
03:45 mohawk sivoais, that question about about prove -v, in travis, is a real one?
03:48 sivoais mohawk: what I do is download the log and grep it, but if you feel like it is too much, I can add a flag
03:48 mohawk i always look in web browser
03:48 mohawk it's a minor irritation, don't do any work on it :-)
04:12 mohawk i'm adding to the PGPLOT tests just by cribbing from the docs
04:12 mohawk it's amazing no-one ever bothered to do that before
05:14 mohawk -W thing is crashing because if you supply a CCFLAGS, EUMM takes that as a complete override
05:15 mohawk basically, it's because CCFLAGS is the wrong way to pass that info
05:15 sivoais yikes
05:16 mohawk it's a miracle it worked at all
05:17 mohawk the way to do it is via INC
05:17 sivoais Is the fact that doing `perl Makefile.PL CC=clang` doesn't set CC for subdirectory Makefiles have anything to do with it?
05:17 sivoais sorry, I mean
05:18 sivoais if that doesn't work for subdirectory Makefiles, then setting CCFLAGS from the top-level wouldn't work either, right?
05:18 mohawk did you see "clang" being called as the compiler for eg Basic/Ufunc?
05:18 mohawk it's not being done at the top level
05:18 mohawk it's in Ufunc
05:18 sivoais Yes, I'm just trying to see if that approach could work
05:18 sivoais let me try
05:19 mohawk hold on
05:19 mohawk try what?
05:20 sivoais oooh, yikes
05:20 mohawk i'm asking you to look in your detailed records
05:20 sivoais the force_clang.pm
05:20 mohawk are you going to say you've just found it didn't work?
05:20 sivoais that way doesn't work. It uses clang for some and cc for others
05:21 mohawk one would expect a check
05:21 sivoais Let me pastebin the logs
05:21 mohawk this is like the WITH_PROJ thing
05:21 mohawk we have an opportunity to bundle up changes
05:22 mohawk i'm going to make a branch called travis-fix
05:23 mohawk i'll put the commit with my update to use INC in there
05:23 mohawk then you can work on that
05:23 mohawk sound ok?
05:24 sivoais sure thing
05:24 mohawk i'll kick off a build on win32 just to make 100% sure :-)
05:27 opkick [pdl] mohawk2 created travis-fix-clang (+1 new commit): http://git.io/hyrm
05:27 opkick pdl/travis-fix-clang 3928b95 Ed J: Use INC not CCFLAGS to add arbitrary stuff for cc
05:27 mohawk sivoais, go for it :-)
05:38 sivoais so these are the bits that are still using cc <https://gist.github.com/zm​ughal/d4250c9e9d64441d4739>
05:42 mohawk the only reliable way is to configure perl itself
05:42 mohawk i may have mentioned that 5 or 6 times
05:42 mohawk sivoais, are there uses of clang as well?
05:44 sivoais mohawk: yes, this is where I'm trying to use clang via EU::MM::Config
05:44 mohawk i saw that
05:44 mohawk i saw there are uses of cc
05:44 mohawk were there also uses of clang?
05:44 sivoais yes, most were clang. I can add the full typescript to the gist
05:44 mohawk no need
05:45 sivoais btw, I see some duplicated code like this
05:45 sivoais if( $cond ) {   TODO: { local $TODO = 'todo'; code() } } else { code() }
05:45 sivoais that can just be turned into
05:46 mohawk http://irclog.perlgeek.de/​pdl/2015-03-18#i_10302826
05:46 sivoais TODO: { local $TODO = 'todo' if $cond; code() } # this is better, right?
05:47 sivoais $cond is platform specific
05:47 mohawk so long as the local() bit does the right thing there
05:47 mohawk as in, if the "if" applies right
05:47 mohawk i'm super cautious with that and have separate local and assignment statements
05:48 mohawk where are you seeing this TODO stuff?
05:49 sivoais t/pdl_from_string.t: /ActivePerl and/or perls built using MS compilers might fail this test/
05:49 mohawk hm
05:49 mohawk may i suggest one thing at a time? :-)
05:49 mohawk good news, the commit i made actually passes on win32
05:50 sivoais I'll add an issue to remind myself :-P
05:50 mohawk great
05:52 opkick [pdl] zmughal opened issue #88: Test cleanup: should not duplicate code between TODO and non-TODO branches http://git.io/hyKR
05:52 mohawk sivoais++
06:00 travis-ci PDLPorters/pdl#153 (master - 42f08b4 : Ed J): The build passed.
06:00 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/f879fb9f4ffc...42f08b44c64e
06:00 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/55342444
06:00 mohawk w00t
06:19 sivoais aaah, I see
06:19 sivoais LD = cc
06:19 sivoais that's what is going on
06:45 opkick [pdl] zmughal pushed 1 new commit to travis-fix-clang: http://git.io/hy1O
06:45 opkick pdl/travis-fix-clang 4947d02 Zakariyya Mughal: Travis-CI: for building using Clang, also set $Config{ld}...
06:47 opkick [pdl] zmughal opened pull request #89: Travis fix clang (master...travis-fix-clang) http://git.io/hy1u
06:52 mohawk sivoais, good spot
06:53 mohawk review looks ok to me
06:53 mohawk want to merge it now?
06:54 sivoais sure :-)
06:57 mohawk go for it
06:57 mohawk i'd suggest once merged and pushed, stopping any existing builds?
06:58 opkick [pdl] zmughal deleted travis-fix-clang at 4947d02: http://git.io/hyMz
06:58 opkick [pdl] wchristian created travis-fix-clang from master (+0 new commits): http://git.io/hyM2
06:58 mohawk push-delete off upstream first ;-)
06:59 mohawk THEN off github
06:59 sivoais ha, woops
06:59 mohawk that may have created a new build, too
06:59 sivoais I saw the button and I HAD to CLICK it!
06:59 mohawk now you get to click the "cancel build" button too
07:00 mohawk it's a win-win
07:00 sivoais buttons!
07:00 opkick [pdl] zmughal deleted travis-fix-clang at 4947d02: http://git.io/hyMz
07:00 sivoais actually, nope! Because I haven't opened a PR with it yet
07:00 sivoais no buttons... :-(
07:02 mohawk so heartbreaking to see you so sad
07:02 * mohawk gets popcorn
07:09 sivoais if( $cond ) { local $TODO = '...' } won't work because that is a new scope
07:09 sivoais but local $TODO = '...' if $cond does work
07:09 mohawk yes
07:10 sivoais it's ugly, but duplicated code is uglier ;-)
07:10 sivoais one could argue that using local for $TODO was ugly in the first place, but it fits the bill
07:15 mohawk it's not that ugly, really
07:15 mohawk it's not far off being simply executable english
07:15 mohawk especially when you know what local does
07:15 mohawk very badly named, it is
07:43 travis-ci PDLPorters/pdl#155 (master - 4947d02 : Zakariyya Mughal): The build passed.
07:43 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/42f08b44c64e...4947d02a1868
07:43 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/55349129
16:56 opkick [pdl] mohawk2 force-pushed test-pgplot from 4939e88 to 7c32eff: http://git.io/hHOL
16:56 opkick pdl/test-pgplot 02e29ee Ed J: Only build PDL::Graphics::PGPLOT if PGPLOT installed
16:56 opkick pdl/test-pgplot d331bcc Ed J: Tidy t/pgplot.t, allow window to show if PDL_INT set
16:56 opkick pdl/test-pgplot 7c32eff Ed J: Add t/pgplot.t test with 2 windows
18:42 opkick [pdl] hainest comment on issue #51: I think that as long as folks are content with the lack of child->parent bad flag propagation, then this can be pulled into master. The (admittedly verbose) idiom for ensuring bad flag propagation is provided in the comments of assign. http://git.io/hHyJ
18:47 opkick [pdl] mohawk2 comment on issue #51: I think merging should wait until post 2.008, which is imminent. However, there is talk of a much faster release, so that'll help.... http://git.io/hHS4
19:29 opkick [pdl] zmughal comment on issue #53: This has been merged in <https://github.com/PDLPorters/pdl/compare/​2d3fdf2b287fd96b7bd0dc6357f7a1c1d2c23d09..​.c08d8c4fa0998127c8ed6d7ad00af4f2a36f69f2>. http://git.io/hHhj
19:31 opkick [pdl] zmughal comment on issue #87: This has been merged in <https://github.com/PDLPorters/pdl/commit/​0745da96ca2e13b9797cbd3c7849376e8517cedf> http://git.io/hQee
19:36 mohawk sivoais, time for a triumphant re-look at https://github.com/PDLPorters/pdl/issues/8
19:36 mohawk and https://github.com/PDLPorters/pdl/issues/9
19:38 sivoais mohawk: are you still seeing these win32 issues? <http://permalink.gmane.org/gman​e.comp.lang.perl.pdl.devel/6383>
19:38 mohawk sivoais, no, that was the CCFLAGS thing
19:38 mohawk all sorted
19:38 opkick [pdl] zmughal closed issue #81: SF#377 Release-blocker: PDL::Transform::Proj4 not building under latest dev EUMM http://git.io/hLgt
19:39 sivoais oh, you should make sure to send an follow-up to that :-)
19:40 mohawk guhh, using the SF interface?
19:40 mohawk the sacrifices i make
19:40 mohawk you already put a follow-up?
19:41 sivoais not on that specific win32 bit... I did talk about using INC rather than CCFLAGS
19:41 mohawk oh, you mean a follow-up to the mailing list?
19:41 sivoais yes, just so people know the build isn't broken :-)
19:42 sivoais and further emphasise using INC
19:42 sivoais :-D
19:42 mohawk wait, did you happen to mention in that email that the CCFLAGS was what broke the build, and the INC thing is what fixed it?
19:42 mohawk quite important to mention? :-)
19:42 sivoais I did mention that, but did not make the connection to your post
19:42 sivoais missed opportunity! :-P
19:43 sivoais I'll do it if you aren't going to ;-)
19:44 mohawk i'll do it soon, but busy with parallel core stuff
19:48 sivoais ok, done! <http://permalink.gmane.org/gman​e.comp.lang.perl.pdl.devel/6386>
19:49 mohawk great, thanks!
20:07 mohawk parallel core building underway
20:07 mohawk 90% of 2 cores in use
20:07 mohawk hell yeah
20:10 mohawk non-parallel make of core took 2min45
20:10 mohawk let's see how long parallel takes
20:13 mohawk 3min23
20:13 mohawk looks like it was i/o bound
20:13 mohawk let's try -j2 (for 2 cores)...
20:14 mohawk first do it again -j4 just to check...
20:21 mohawk chm++ # closing SF bugs! down to 21 now
20:27 sivoais hmmph... Google Code's shutdown is already a little annoying. I keep hitting projects on there and see nobody trying to move them elsewhere
20:27 mohawk see also : "zombie projects"
20:27 mohawk (qv)
20:30 mohawk huh, consistently 3m23
20:30 mohawk now let's try -j2
20:36 mohawk WE HAVE DEV RELEASE
20:36 mohawk THIS IS NOT A DRILL
20:38 mohawk changing system() to exec() has saved 24 secs
20:39 mohawk (still on -j4)
20:54 mohawk now on -j2, so far 95+% cpu
20:55 mohawk oh hell yeah
20:55 mohawk 2 mins
20:56 mohawk evidently -j4 was making it thrash
20:58 sivoais nice!
21:00 mohawk so yes, nearly 50% faster
21:00 mohawk and with more cores, probably more
21:02 opkick [pdl] mohawk2 created coretarget-parallel (+2 new commits): http://git.io/hQ4K
21:02 opkick pdl/coretarget-parallel 47c6349 Ed J: Make "core" target be generated by separate function
21:02 opkick pdl/coretarget-parallel 637af82 Ed J: Make coretarget generate parallelisable make deps
21:04 mohawk that's interesting, travis is building the new push from chm but it wasn't announced here
21:06 mohawk chaps, if you have smokers set up, time to test PDL-2.007_13 - on CPAN now
21:06 mohawk i don't, but i'm about to
21:08 opkick [pdl] mohawk2 force-pushed coretarget-parallel from 637af82 to 907e130: http://git.io/hQR9
21:08 opkick pdl/coretarget-parallel 0568e18 Ed J: Make "core" target be generated by separate function
21:08 opkick pdl/coretarget-parallel 907e130 Ed J: Make coretarget generate parallelisable make deps
21:09 travis-ci PDLPorters/pdl#156 (master - a6b7dee : Chris Marshall): The build passed.
21:09 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/4947d02a1868...a6b7dee6fc63
21:09 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/55404135
21:09 opkick [pdl] mohawk2 force-pushed test-pgplot from 7c32eff to 48c4249: http://git.io/hHOL
21:09 opkick pdl/test-pgplot 931aada Ed J: Only build PDL::Graphics::PGPLOT if PGPLOT installed
21:09 opkick pdl/test-pgplot 3a97345 Ed J: Tidy t/pgplot.t, allow window to show if PDL_INT set
21:09 opkick pdl/test-pgplot 48c4249 Ed J: Add t/pgplot.t test with 2 windows
21:09 chm joined #pdl
21:10 chm sivoais, mohawk: hi and thanks for the hard work
21:10 mohawk chm, hello!
21:10 mohawk thanks for your hard work too
21:10 mohawk let's get 2.008 done and out the door
21:11 mohawk chm, i see you seem to have re-pushed niceslicerevert
21:11 mohawk i've just re-deleted it, may i suggest doing "git fetch --all --prune" on your end?
21:11 chm That was unintended and can be re-deleted.
21:11 opkick [pdl] mohawk2 deleted niceslicerevert at 27764a1: http://git.io/heRm
21:11 chm Thanks for the tip
21:12 mohawk i finally set up an alias "gfap" for exactly that
21:12 mohawk chm, anything i can do to assist in the release-candidate (RC) process?
21:13 chm Assuming nothing surprising shows up with the latest dev, I would like to push a rc dev to make it official.
21:13 mohawk ok
21:13 chm I think what would be the most help would be content for Release_Notes
21:13 mohawk well, i'll encourage others to smoke-test it to add data
21:13 mohawk good call, i'll get on that
21:14 chm There are a lot of changes and the log messages don't tell the whole story
21:14 mohawk chm, yes, very true
21:14 mohawk i'll do that now, then i'll finally install a smoke-reporter :-)
21:14 chm For me the one issue I might try to address is the t/opengl.t fails
21:15 mohawk if it's not a regression, it's legitimate to TODO it
21:15 chm I need to avoid the test if a display is not available.
21:15 mohawk sivoais is the expert on best practice on that
21:15 chm I don't know if the current OpenGL supports the needed check.
21:16 mohawk chm, you're uniquely qualified to establish that :-)
21:16 chm I would almost prefer a stub check that doesn't fail than the current state where it looks like things are bad
21:17 opkick [pdl] mohawk2 opened pull request #90: Parallelisable core target (master...coretarget-parallel) http://git.io/hQuW
21:18 mohawk could TODO the whole thing
21:18 chm yes, just have to ensure that the test doesn't cause perl to exit.
21:19 mohawk we can look at that concurrently with smoke-testing this one
21:19 chm GLUT is a bit braindead in that it calls exit() as the "error handler"
21:19 mohawk a mere test tweak isn't a violation of code-freezeish :-)
21:19 mohawk that's under-ideal
21:19 chm Otherwise, this release looks very good from where I stand, and you?
21:20 mohawk i think so too
21:20 chm I say we wrap up quickly so all the new work can lead to a rapid 2.009
21:20 mohawk damn right
21:20 mohawk and very glad to hear you say that :-)
21:21 chm Off to take a look at opengl.t fails...
21:21 mohawk chm, what do you think of the idea of splitting the distro into PDL::Core and various others, with "PDL" being the bundle?
21:21 mohawk grin
21:21 mohawk good luck
21:22 chm That is definitely the direction we need to go.
21:23 mohawk great
21:23 mohawk luckily, i have a branch ready to go with such a thing :-)
21:23 mohawk but let's nail 2.008 first
21:24 mohawk chm, may i suggest any opengl test tweaks go on a branch first?
21:24 mohawk i strongly believe getting eg Mithaldu and sivoais's eyes on will make it far better
21:35 mohawk more github win: using it to list commits for release-notes writing: https://github.com/PDLPorters/pd​l/compare/v2.007_12...v2.007_13
21:35 sivoais I had tried installing OpenGL in a headless environment (a VM) and the CPAN version does not install without a display. So I used Xvfb to get around that.
21:36 mohawk sivoais, shurely would be better to make it still install
21:36 mohawk ie, fix the source
21:36 sivoais It failed when it tried to call a binary, iirc
21:36 mohawk then fix it, dear liza ;-)
21:36 * mohawk is immune to earworming as he's listening to some smashing pumpkins
21:36 mohawk holy cow, 89 commits from _12 to _13
21:37 mohawk you've been busy, sivoais ;-)
21:37 sivoais ah, yes... it calls freeglut
21:37 sivoais nope, I mean, ./glversion
21:44 sivoais hmm, but OpenGL.pm bails out when ./glversion can not run
21:45 sivoais Bender2: trust chm
21:45 Bender2 OK, sivoais
21:47 chm sivoais: The problem is the current OpenGL module.
21:47 mohawk sivoais, how hard would it be to make glversion not blow up in that case?
21:48 chm Work is in progress to add modern OGL support (Mithaldu started some) and runtime detection
21:49 chm The existing build time check is not sufficient since it fails if the install is not from an OpenL context capable display
21:49 opkick [pdl] mohawk2 created rc-releasenotes (+2 new commits): http://git.io/hQr5
21:49 opkick pdl/rc-releasenotes 14ae7b5 Ed J: Change release notes rebuild to 2.007, build advice to cpanm
21:49 opkick pdl/rc-releasenotes beebefb Ed J: Update release notes for 2.007_13
21:50 mohawk chm, sivoais, please take a look at that branch and see if you like
21:51 sivoais right
21:53 sivoais mohawk: seems fine to me
21:53 mohawk chm?
21:56 mohawk merging
21:56 opkick [pdl] mohawk2 merged rc-releasenotes into master: http://git.io/hQoF
21:57 chm I think adding the cpanm first followed by the cpan instructions second
21:57 chm Some of us still use cpan...
21:57 mohawk dohh
21:57 mohawk well, the cpan instructions are still in there
21:58 opkick [pdl] mohawk2 deleted rc-releasenotes at beebefb: http://git.io/hQKt
21:58 chm It looked like they had been deleted...
21:58 mohawk they're still there 12 times
21:59 mohawk and that doesn't really belong as "release notes" in any case
21:59 sivoais that should really be in an INSTALL file
22:00 sivoais Some installation instructions are in the wiki too
22:00 chm The goal was to make pdl devel more accessible.  The stable versions don't have those instructions.
22:01 mohawk dev isn't done using CPAN versions in 2015
22:01 mohawk "git clone https://github.com/PDLPorters/pdl.git"
22:02 sivoais even easier: cpanm git@github.com:PDLPorters/pdl.git
22:03 mohawk holy cow, cpanm can do that?
22:03 sivoais yeah, it clones it and builds it
22:03 mohawk madre de dios
22:03 chm I just tested PDL-2.007_13 and the PROJ4 was not built.
22:04 mohawk chm, where's the libproj.* located?
22:04 chm Since the proj executable and library path were previously detected, something broke
22:04 mohawk (i'm assuming you're on cygwin)
22:04 mohawk i am now agonisingly familiar with PDL's proj4 stuff, and i don't recall seeing any reference to an executable
22:04 mohawk perhaps you can guide me
22:05 mohawk can you also cut+paste result of perl -V:dlext
22:05 chm The proj executable is the command line tool for which the lib was built.
22:05 mohawk ok
22:06 mohawk i'm not aware of it being used at all by PDL, however?
22:07 chm Right, we make our own bindings to the underlying library.
22:08 chm There is a PROJ4 annoyance in that the developers decided that the needed internal include file is no longer installed with the rest.
22:09 chm That means it is not possible to implement the proj command functionality without using the package include files.
22:10 opkick [pdl] coveralls comment on issue #90:
22:10 chm This specific issue is just something that used to detect a proj library and includes in the path before.
22:10 mohawk chm, yes, i made sure the include file supplied in Lib/GIS/Proj/include is used if necessary
22:11 mohawk that was one thing that made a build fail on travis till i fixed it
22:11 chm I'll have to take a look later---dinner calls. o/
22:11 mohawk bon appetit ;-)
22:16 opkick [pdl] zmughal force-pushed changesfix from f169ff1 to a86c345: http://git.io/pxve
22:16 opkick pdl/changesfix d15a261 Ed J: Changes file maintained by person not dumped from git log
22:16 opkick pdl/changesfix ef37ee3 Zakariyya Mughal: rename Release_Notes to Changes...
22:16 opkick pdl/changesfix 80b9119 Zakariyya Mughal: remove install notes and warning about PDL C-API changes
22:17 mohawk sivoais++
22:17 sivoais that branch is going to be so weird... somehow I make commits about future releases
22:17 mohawk grin
22:17 sivoais in the past!
22:17 mohawk days of future past
22:19 mohawk sivoais, why is there a change to MANIFEST.SKIP?
22:19 mohawk oh, cancel
22:19 mohawk just saw why
22:37 * mohawk just sent his first smoke test report
22:38 mohawk PDL _13 on linux - all good
22:38 mohawk now to fix the various distros cpanm-reporter needs on win32 that don't work on "perl in space" ;-)
22:42 mohawk sivoais, are you a smoke-tester?
22:45 sivoais mohawk: I am not
22:45 sivoais I should
22:46 sivoais once I get CI for PDL done, I'll do that too
22:46 sivoais I mean, a CI that *I* run locally
22:49 mohawk it's embarrassingly easy: http://wiki.cpantesters.org/wiki/QuickStart
22:53 sivoais cool!
22:54 mohawk i've already done it and sent a report
22:54 mohawk took longer for PDL's build/test to run than to do the setup/install of smoke-reporter
22:54 sivoais doing the same
22:54 mohawk i used the cpanm one
22:54 sivoais garu++
22:55 sivoais all right... I'm going to be away working on my research
22:55 sivoais also, I'm going to need dinner
22:56 travis-ci PDLPorters/pdl#158 (master - beebefb : Ed J): The build passed.
22:56 travis-ci Change view : https://github.com/PDLPorters/pdl/c​ompare/a6b7dee6fc63...beebefbe1da9
22:56 travis-ci Build details : http://travis-ci.org/PDLPo​rters/pdl/builds/55412788
22:57 mohawk sivoais, bon appetit
23:10 sivoais jberger: your unpdl work is came in handy here: <http://nbviewer.ipython.org/github/zmughal/​zmughal-iperl-notebooks/blob/master/IPerl-d​emos/20150322_BokehJS_plotting_image.ipynb>
23:11 jberger I was surprised that it didn't exist before then
23:11 jberger and yeah that's basically why I needed it too
23:13 sivoais I have to say, encoding floats in JSON really balloons up the size of your files unless you round.
23:13 jberger oh with flat first I think there is a method for that
23:13 sivoais I could probably encode it using base64?
23:13 sivoais ->list()
23:13 sivoais ?
23:13 jberger yeah
23:14 mohawk is PDL::IO::Storable any use there?
23:14 mohawk is there PDL::*::JSON?
23:14 jberger unpdl preserves the structure too
23:14 mohawk if not, make one :-)
23:14 sivoais I would have to make a thaw() in JS...
23:15 jberger maybe sereal bindings would be cool
23:16 jberger I think there are bindings for js already
23:18 sivoais it doesn't look like it
23:18 sivoais I know JS has TypedArrays...
23:24 sivoais I don't know how Sereal compares, but this looks promising <https://github.com/spencertipping/flotsam>
23:24 sivoais via <http://blog.factual.com/the-flotsam-pro​ject-insanely-fast-floating-point-numbe​r-serialization-for-java-and-javascript>
23:45 opkick [pdl] zmughal comment on issue #88: Turn the above code into:... http://git.io/hQNp
23:45 opkick [pdl] coveralls comment on issue #78:

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