Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2015-08-29

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

All times shown according to UTC.

Time Nick Message
00:27 opkick [PDL-Stats] mohawk2 deleted kmeans-test-update at 315b975: http://git.io/vGstP
00:31 opkick [PDL-Stats] mohawk2 commented on issue #6: I'm going to rebase this branch against latest master and force-push. That will rerun the build test. http://git.io/vGstd
00:31 opkick [PDL-Stats] mohawk2 commented on issue #6: In fact I'm not, since that's already been done. Since these changes are human-orientated only, I'm going to merge them now and then *that* will rerun the build! http://git.io/vGsqL
00:31 opkick [PDL-Stats] mohawk2 pushed 1 new commit to master: http://git.io/vGsqt
00:31 opkick PDL-Stats/master 08cf4a4 mohawk2: Merge pull request #6 from PDLPorters/gh-sprucing...
00:31 opkick [PDL-Stats] mohawk2 deleted gh-sprucing at 71405c8: http://git.io/vGsqq
00:47 travis-ci PDLPorters/PDL-Stats#11 (master - 08cf4a4 : mohawk2): The build was fixed.
00:47 travis-ci Change view : https://github.com/PDLPorters/PDL-Stats/compare/00054aed5ce1...08cf4a4aa9b6
00:47 travis-ci Build details : https://travis-ci.org/PDLPorters/PDL-Stats/builds/77772061
00:48 opkick [PDL-Stats] mohawk2 created todo-nondeterministic (+1 new commit): http://git.io/vGsYM
00:48 opkick PDL-Stats/todo-nondeterministic 806a294 Ed J: Mark non-deterministic test TODO as fails on clang
00:50 opkick [PDL-Stats] mohawk2 opened pull request #10: Mark non-deterministic test TODO as fails on clang (master...todo-nondeterministic) http://git.io/vGsYb
01:02 mohawk sivoais, is pdla enchanted?
01:06 opkick [PDL-Stats] mohawk2 created add-travis-deps (+1 new commit): http://git.io/vGssH
01:06 opkick PDL-Stats/add-travis-deps 3f44532 Ed J: Beef up CI with optional requirements spelled out and apt installed
01:07 opkick [PDL-Stats] mohawk2 opened pull request #11: Beef up CI with optional requirements spelled out and apt installed (master...add-travis-deps) http://git.io/vGssb
01:11 opkick [pdla] mohawk2 commented on issue #3: Considered. No. http://git.io/vGsGz
01:28 sivoais mohawk: as you can see, it is... :-P
01:28 mohawk ha ha, indeed!
01:28 mohawk magic!
01:29 mohawk i'm just starting on converting PDL -> PDLA
01:29 mohawk time estimated: 6 minutes
01:30 sivoais cool. making a record of the commands you used? perl -pi -e "PERL!" $(find -type f -name '*.pm) ?
01:30 mohawk yes
01:30 mohawk so far it's:
01:30 mohawk 19737  vi Basic/Bad/Makefile.PL
01:30 mohawk 19738  vi Basic/Bad/bad.pd
01:30 mohawk 19739  vi Basic/Complex/Makefile.PL
01:30 mohawk 19740  vi Basic/Complex/complex.pd
01:30 mohawk i'll keep you updated as i go ;-)
01:31 mohawk (can't automate it as i'm leaving the C stuff alone, and perldl.conf)
01:31 sivoais yeah, that part I felt would be tricky since that's the ABI
01:32 mohawk there isn't a hope in hell of dodging recompiling other stuff
01:33 mohawk since the purpose is just to make a new PAUSE namespace rather than change everything for the sake of it, i'm going with this strategy: only change perl package stuff (inc docs), for now
01:33 mohawk Bad has a fair old bit of repetition
01:33 mohawk that needs looking at properly
01:33 sivoais post-split, I'd say
01:34 sivoais it's much easier to look at each bit of code if it can be refactored independently of the rest
01:34 mohawk i know ;-)
01:34 mohawk and i agree
01:35 mohawk i assess the split will happen tomorrow
01:35 mohawk and there's no reason not to have PIG and PDLA-Stats done then too
01:35 mohawk if you're around, we can sharpen up devops
01:35 mohawk i'd like you to review the 2 PDL-S PRs if you have time
01:36 sivoais mmm, not tonight. I'm cooking up dinner and then heading back to the lab for a good old-fashioned grind :-)
01:36 mohawk nerd
01:37 sivoais :-P
01:37 mohawk the PRs are quite small, but fair enough
01:37 sivoais I saw the essense of the PRs and they seemed good
01:37 mohawk cool
01:37 mohawk well, not a particular rush
01:37 mohawk i may have underestimated with my first stab of 6 minutes
01:37 sivoais <https://github.com/PDLPorters/PDL-Stats/pull/11> and and <https://github.com/PDLPorters/PDL-Stats/pull/10>?
01:38 mohawk yup
01:39 mohawk what the criminy is this for:
01:39 mohawk L<PDLA::Graphics::PGPLOT|PDLA::Graphics::PGPLOT>
01:39 mohawk (ignore the A)
01:39 sivoais lol
01:39 opkick [PDL-Stats] zmughal commented on issue #10: This seems sensible for now. Since we still have #7 open, it will be there to remind us without annoyingly failing the whole build. http://git.io/vGscY
01:41 mohawk git grep reckons "PDL" occur in 622 files
01:43 mohawk i think i'm going to have to bite the bullet and do a mass change after all
01:45 mohawk this could take a very long time
01:45 mohawk and i already got enough RSI grinding through the awful Makefile.PLs
01:45 opkick [PDL-Stats] zmughal commented on issue #11: Ah, yes, this is so that it will pull in the right modules later on when the PDL split is complete? I say this is worth it.... http://git.io/vGscy
01:45 sivoais mohawk: vi macros!
01:45 mohawk even with those...
01:45 mohawk that means the ABI will change too
01:45 * sivoais bashes @@ repeatedly
01:46 * mohawk prays that's not a euphemism
01:47 sivoais hahah
01:47 sivoais @@ repeats the previous macro
01:48 sivoais so I often just stick a <modification><search for next instance> in a macro and go off to the races
01:49 mohawk oh yes: type vi commands, 0"aD, @a
01:49 mohawk never really used that functionality much
01:52 mohawk perl -pi -e 's#PDL([^A])#PDLA$1#g' $(gg -l PDL)
01:52 mohawk this means that we'll refer to the data objects as PDLAs
01:52 mohawk which makes about as much sense as calling them PDLs
01:53 mohawk now the tedious business of actually reviewing the changes it made!
01:54 opkick [PDL-Stats] mohawk2 deleted todo-nondeterministic at 806a294: http://git.io/vGsCM
01:57 opkick [PDL-Stats] mohawk2 force-pushed add-travis-deps from 3f44532 to 556fd76: http://git.io/vGsWt
01:57 opkick PDL-Stats/add-travis-deps 556fd76 Ed J: Beef up CI with optional requirements spelled out and apt installed
02:05 opkick [PDL-Stats] mohawk2 pushed 1 new commit to add-travis-deps: http://git.io/vGslm
02:05 opkick PDL-Stats/add-travis-deps e55deaf Ed J: PGPLOT verbose install
02:07 opkick [PDL-Stats] mohawk2 force-pushed add-travis-deps from e55deaf to 2425957: http://git.io/vGsWt
02:07 opkick PDL-Stats/add-travis-deps 2425957 Ed J: Beef up CI with optional requirements spelled out and apt installed
02:34 opkick [PDL-Stats] maggiexyz commented on issue #10: Thanks! I'll try to update the test again in the next few days. http://git.io/vGsBi
03:08 opkick [PDL-Stats] mohawk2 pushed 1 new commit to master: http://git.io/vGsuc
03:08 opkick PDL-Stats/master 58c48de mohawk2: Merge pull request #11 from PDLPorters/add-travis-deps...
03:08 opkick [PDL-Stats] mohawk2 deleted add-travis-deps at 2425957: http://git.io/vGsuC
03:10 sivoais mohawk: can those deps be made optional?
03:10 opkick [PDL-Stats] mohawk2 pushed 1 new commit to master: http://git.io/vGsuz
03:10 opkick PDL-Stats/master cc144dc Ed J: v0.71
03:10 opkick [PDL-Stats] mohawk2 tagged v0.71 at master: http://git.io/vGsug
03:11 sivoais because what if someone doesn't have pgplot installed
03:11 mohawk then either their cpan client will install it, or it will warn during configure time
03:11 mohawk i haven't removed the fault tolerance code
03:12 sivoais ah, ok, so it won't bail out on them entirely
03:12 mohawk no
03:15 mohawk wow, chm has some funky "modules" that PDL allegedly supplies
03:16 mohawk C::StructObj C::StructType C::Type C::Var Inline::MakePdlppInstallable Inline::Pdlpp SymTab Win32::DDE::Netscape XS
03:16 mohawk yes, XS
03:17 mohawk so PDLA got indexing failures
03:17 mohawk the tests don't all pass yet, but too bad
03:17 mohawk i'll push the branch in its current state now
03:18 sivoais ok, fine. I'll take a look when I have tuits
03:18 mohawk not too necessary unless you want to, for now
03:18 opkick [pdla] mohawk2 created PDLA (+1 new commit): http://git.io/vGszC
03:18 opkick pdla/PDLA eb031e8 Ed J: PDL->PDLA with: perl -pi -e 's#PDL([^A])#PDLA$1#g' $(gg -l PDL)
03:18 mohawk the FITS code broke because it relies on byte-alignment and "PDLA" is a different number of bytes from "PDL"
03:19 mohawk there's other weirdness but should be straightforward
03:19 mohawk night ;-)
03:19 mohawk (my strategy is to build "real" pdl and see what it's getting where pdla blows up)
03:32 travis-ci PDLPorters/PDL-Stats#18 (master - 58c48de : mohawk2): The build passed.
03:32 travis-ci Change view : https://github.com/PDLPorters/PDL-Stats/compare/166b1e2d1115...58c48de1b98f
03:32 travis-ci Build details : https://travis-ci.org/PDLPorters/PDL-Stats/builds/77780717
03:53 travis-ci PDLPorters/PDL-Stats#19 (master - cc144dc : Ed J): The build passed.
03:53 travis-ci Change view : https://github.com/PDLPorters/PDL-Stats/compare/58c48de1b98f...cc144dc1133e
03:53 travis-ci Build details : https://travis-ci.org/PDLPorters/PDL-Stats/builds/77780779
04:15 jberger joined #pdl
11:21 mohawk sivoais, concurrently while i'm wrestling the PDLA ourobouros into submission, could i ask you to have a hard look at the devops stuff to make sure it's idiot-proof to run, and/or states all its requirements?
11:22 mohawk we're going to be doing a lot of enchanting :-)
11:26 opkick [pdla] mohawk2 opened issue #4: Eliminate modules PAUSE thinks I want to index http://git.io/vGGEo
11:29 opkick [pdla] mohawk2 commented on issue #3: Issues dealt with so far: had to rename a few files/dirs from `PDL*` to `PDLA*`.... http://git.io/vGGuq
12:57 opkick [pdla] mohawk2 commented on issue #3: Currently-failing tests:... http://git.io/vGGSA
12:57 mohawk (including any configuration files and whatnot)
13:16 mohawk sivoais++ # the better tests are making this much easier
13:35 mohawk SV **svp = hv_fetch(hash,"PDLA",3,0);
13:35 mohawk can anyone spot the ass_u_mption made there?
13:43 mohawk (hint: the correct perl function there is hv_fetchs, not hv_fetch - introduced in perl 5.8.9 or so)
13:45 mohawk still, since that line has probably been in there since god was a boy, i'm not really judging
14:02 opkick [pdla] mohawk2 commented on issue #3: The subclassing ones were solved by switching `Basic/Core/pdlcore.c.PL` to use `hv_fetchs` instead of `hv_fetch`. http://git.io/vGGhX
14:37 opkick [pdla] mohawk2 commented on issue #3: Storable one was due to supplied test data files having a leading length byte before "PDLA" not having been updated, plus space-padding after `PDLA_B` needing adjustment. http://git.io/vGZTK
14:56 opkick [pdla] mohawk2 commented on issue #3: `proj_transform.t` one was also a data alignment problem: `Lib/Transform/Cartography/earth_coast.vec.fits`.... http://git.io/vGZme
15:08 opkick [pdla] mohawk2 commented on issue #3: @devel-chm I'm quite enjoying the irony of you being against polluting a top-level namespace while PDL claims to provide a module called (and this is its full name): `XS`. https://metacpan.org/source/CHM/PDL-2.013/Basic/Gen/PP/XS.pm... http://git.io/vGZOz
15:15 opkick [pdla] mohawk2 opened pull request #5: PDL->PDLA (master...PDLA) http://git.io/vGZ3y
15:15 mohawk sivoais, jberger - review sought for PR #5
15:15 mohawk i'm going to branch off it for the "stop claiming all those modules" effort
15:16 opkick [pdla] mohawk2 force-pushed PDLA from eb031e8 to 80784e9: http://git.io/vGZsc
15:16 opkick pdla/PDLA 49ea964 Ed J: No hardcode length of HV key
15:16 opkick pdla/PDLA 80784e9 Ed J: PDL->PDLA with: perl -pi -e 's#PDL([^A])#PDLA$1#g' $(gg -l PDL)
15:17 mohawk sivoais, it would probably be useful if you could stop this build: https://ci.appveyor.com/project/zmughal/pdla/build/1.0.10
15:18 mohawk and also, would it be possible to authorise others to do so?
15:25 opkick [pdla] devel-chm commented on issue #3: Ed, I have co-maint or better for all PDL dist so I can make releases.  And
15:32 sivoais mohawk: I think I have given perms on it to the GitHub team now... Appveyor's model is a little confusing
15:32 mohawk ha ha, ok
15:33 mohawk any chance of a review of PR #5?
15:33 mohawk you're probably best off just reading the comments and giving a hyperskimmy skim to the code changes
15:33 sivoais yeah, let me put some tea on ;-)
15:33 mohawk given there's quite a bit there
15:33 mohawk grin
15:33 mohawk i'm just going to fixup the packages stuff then it's new release time
15:36 opkick [pdla] mohawk2 commented on issue #3: It's a shame that all that proponent-ing hasn't lead to any identifiable activity in the last year. http://git.io/vGZcs
15:44 mohawk TIL that it's all very well to modify a file in your editor, but it's not going to help as much as it could if you don't save it to disk
15:52 sivoais haha
15:56 mohawk check this out: what do you think Basic/Gen/PP/Struct.pm does?
15:56 mohawk clue: nothing at all, whatsoever
15:56 mohawk ...except claim 2 namespaces on PAUSE
15:56 mohawk *zot*
16:01 mohawk sivoais, when you've reviewed #5 could you do a squash activity on pdla's tc3? i want to convert and merge it
16:23 opkick [pdla] zmughal commented on issue #5: To get a readable view on the diff between the branches, run... http://git.io/vGZuh
16:38 opkick [pdla] mohawk2 created fixpackageclaim (+8 new commits): http://git.io/vGZVY
16:38 opkick pdla/fixpackageclaim 61740d8 Ed J: s/XS/PDLA::PP::XS/
16:38 opkick pdla/fixpackageclaim 829be0a Ed J: Hide Win32::DDE::Netscape from PAUSE
16:38 opkick pdla/fixpackageclaim 6ec8232 Ed J: s/SymTab/PDLA::PP::SymTab/
16:39 opkick [pdla] zmughal commented on issue #5: Looks fine for now. I'd like to get this completely done before moving forward with any other changes. It will make it easier to reason about any breakage in other branches. http://git.io/vGZV2
16:40 opkick [pdla] mohawk2 opened pull request #6: Fix spurious packages claim (master...fixpackageclaim) http://git.io/vGZVX
16:40 mohawk sivoais, i'm going to merge #5
16:40 mohawk please review #6
16:40 mohawk when that's merged, i'll release again to check packages issues
16:41 mohawk once that's done, then we split
16:42 sivoais mohawk: but does #5 pass its build yet?
16:42 mohawk no
16:42 mohawk because t/lut.t is fragile
16:42 mohawk i TODO-ed it in #6
16:42 mohawk therefore, i propose we ignore that for now
16:42 mohawk work for you?
16:43 sivoais ok, sure
16:43 mohawk if you're content, please comment on #5
16:44 sivoais is #6 on top of #5?
16:44 mohawk yes
16:45 mohawk since i'm planning to make and release a split today, i want to advocate a temporarily casual attitude to build passing (though not to review and code quality)
16:46 sivoais make sense
16:46 mohawk am i right you're ok with merging #5?
16:46 mohawk i want to tag it as being fairly close to the 2.013 i put out this morning
16:47 opkick [pdla] zmughal commented on issue #6: This looks ready to merge after #5 + rebase.... http://git.io/vGZrW
16:47 mohawk nice
16:47 opkick [pdla] zmughal commented on issue #5: **Merge**: *Aye* http://git.io/vGZrE
16:48 opkick [pdla] mohawk2 merged PDLA into master: http://git.io/vGZri
16:49 opkick [pdla] mohawk2 tagged pdla2.013 at PDLA: http://git.io/vGZr5
16:51 opkick [pdla] mohawk2 deleted PDLA at 80784e9: http://git.io/vGZoO
16:52 opkick [pdla] mohawk2 opened issue #8: Appveyor and README.md need attention to PDLA-ise them right http://git.io/vGZow
16:54 sivoais right
16:54 sivoais squashing tc3 now
16:54 sivoais going to PDLA-ise the patch
16:55 opkick [pdla] mohawk2 commented on issue #6: No rebase necessary as I branched it off there already. Next step is release 2.013001 http://git.io/vGZK3
16:55 mohawk great
16:55 mohawk let's include that in 2.013001 then
16:55 mohawk actually no
16:55 mohawk soon as travis finishes (ETA 2 mins) i'll release 001
16:55 mohawk that allows us to see any PAUSE issues while you sort tc3
16:55 mohawk and i can go do some exercise like a keen boy
16:56 mohawk ...unless you're keen and have already PRed it ;-)
16:57 mohawk which you have, but it's not squashed yet which is fine
16:57 mohawk if it's ready, i'll include it
16:57 mohawk if not, later today
16:57 mohawk but pre split so the split catches all the file-moving
16:58 mohawk green light on #6
16:59 opkick [pdla] mohawk2 deleted fixpackageclaim at 8d6584d: http://git.io/vGZ6C
17:00 mohawk it is SO nice not having to push to SF *and* to gh
17:02 sivoais I had added multiple push remotes to my origin :-P
17:02 opkick [pdla] mohawk2 pushed 1 new commit to master: http://git.io/vGZin
17:02 opkick pdla/master 4a6383b Ed J: 2.013001
17:02 mohawk i contemplated it
17:02 mohawk but i found a better solution
17:02 opkick [pdla] mohawk2 tagged pdla2.013001 at master: http://git.io/vGZiB
17:05 mohawk 2.013001 on PAUSE now
17:10 opkick [pdla] mohawk2 opened issue #9: Makefile.PL: remove pure_nolink, also simplify coretarget http://git.io/vGZXG
17:11 travis-ci PDLPorters/pdla#6 (master - 80784e9 : Ed J): The build was broken.
17:11 travis-ci Change view : https://github.com/PDLPorters/pdla/compare/b26afaaee1b6...80784e9d2ccb
17:11 travis-ci Build details : https://travis-ci.org/PDLPorters/pdla/builds/77831725
17:12 mohawk don't panic
17:12 mohawk that's the one i should have terminated
17:16 mohawk Status of this distro: OK
17:16 mohawk GET IN
17:17 mohawk this is what "release early, release often" looks like, people
17:17 mohawk #justsaying
17:19 mohawk i'm just fixing up the "coretarget" changes while i wait for tc3
17:23 mohawk ok, this multi-place $VERSION is next on my hitlist
17:25 travis-ci PDLPorters/pdla#7 (master - 8d6584d : Ed J): The build passed.
17:25 travis-ci Change view : https://github.com/PDLPorters/pdla/compare/80784e9d2ccb...8d6584d4fbea
17:25 travis-ci Build details : https://travis-ci.org/PDLPorters/pdla/builds/77832859
17:25 mohawk there ya go
17:26 sivoais wooo
17:27 sivoais mohawk++
17:27 mohawk ha ha, thanks
17:27 sivoais tc3 is almost there. Have to resolve the lut.t changes
17:27 mohawk i think 2.013001 won't work
17:27 mohawk yeah, sorry about that
17:27 mohawk because it has 2 $VERSIONs one in Basic/PDLA.pm, one in Basic/Core/Core.pm
17:28 mohawk i'm going to make it have just one
17:28 sivoais no problem, it's done! squash + push now
17:29 mohawk magic
17:29 opkick [pdla] zmughal force-pushed test-cleanup-3 from d867aae to 33bb71e: http://git.io/vGZyt
17:29 opkick pdla/test-cleanup-3 33bb71e Zakariyya Mughal: test cleanup part 3...
17:29 mohawk i'll give a quick review now
17:30 mohawk suggest you terminate build that's in progress
17:30 mohawk can you draw my attention to anything that i didn't previously comment about?
17:31 sivoais just lut.t. It's the same as before, just running perl -pi -e on the patch
17:31 opkick [pdla] mohawk2 opened issue #10: Revert Changes* PDLA->PDL, update http://git.io/vGZyr
17:31 mohawk did you fix up the things i commented about?
17:31 sivoais yep
17:31 sivoais a while back
17:31 mohawk good enough, then
17:32 mohawk merge away
17:32 opkick [pdla] mohawk2 commented on issue #7: Great work! Aye. http://git.io/vGZSO
17:33 sivoais waiting for the CIs
17:35 mohawk great
17:43 opkick [pdla] mohawk2 force-pushed coretarget-parallel from 49ce096 to e06ccad: http://git.io/vGZQm
17:43 opkick pdla/coretarget-parallel d5bc88b Ed J: Make "core" target be generated by separate function
17:43 opkick pdla/coretarget-parallel f48f724 Ed J: Make coretarget generate parallelisable make deps
17:43 opkick pdla/coretarget-parallel e06ccad Ed J: Basic/Core/Core.pm get $VERSION from PDLA.pm
17:47 mohawk sivoais, if you're at a loose end, could you attack #10 please?
17:48 mohawk could you also make appveyor's build rule match travis, if that's possible?
17:50 sivoais yes to both in a bit
17:52 mohawk awesome!
17:58 mohawk sivoais, i just gave you co-maint on the approx 600,000 modules in PDLA
17:59 mohawk ok, i lied
17:59 mohawk only 382
18:01 sivoais hahah
18:01 sivoais that scared me for a bit... sounds like 1/2 of CPAN
18:01 mohawk uh.... you're welcome?
18:01 opkick [pdla] mohawk2 force-pushed coretarget-parallel from e06ccad to 100b7d5: http://git.io/vGZQm
18:01 opkick pdla/coretarget-parallel 57865cb Ed J: Basic/Core/Core.pm get $VERSION from PDLA.pm
18:01 opkick pdla/coretarget-parallel 58caafe Ed J: Make "core" target be generated by separate function
18:01 opkick pdla/coretarget-parallel eedc2e0 Ed J: Make coretarget generate parallelisable make deps
18:03 sivoais I'll get to those pdla issues once I figure out this linker problem I'm having in my code.
18:03 sivoais I'm trying to get a huge C++ library to link with my C code
18:05 mohawk uh oh
18:05 mohawk can i help?
18:08 opkick [pdla] mohawk2 pushed 1 new commit to coretarget-parallel: http://git.io/vGZbh
18:08 opkick pdla/coretarget-parallel 1821b3c Ed J: Eliminate pure_nolink
18:09 opkick [pdla] mohawk2 opened pull request #11: Coretarget parallel (master...coretarget-parallel) http://git.io/vGZNT
18:09 sivoais hmm, let me see if I can package up the work so far. The biggest issue with work with this library is that it has no pkg-config equivalent. It wraps up everything using Cmake and I avowedly avoid Cmake when I can
18:09 mohawk sivoais, a superquick review of #11 would be useful
18:09 mohawk yow, that sounds challenging
18:09 mohawk are there any tools you could use?
18:10 sivoais looking at #11
18:11 sivoais there is one thing I've been trying which is to get CMake to dump out it's LDLIBS. It almost works, but get this... it gives me the -l$libname for everything, but the actual files on disk are lib$libname-$version.so
18:11 sivoais but not all of them
18:12 sivoais /o\
18:12 opkick [pdla] mohawk2 created changesfix (+1 new commit): http://git.io/vGZNy
18:12 opkick pdla/changesfix 38cec72 Ed J: Revert accidental historical revisionism
18:12 mohawk oh wow
18:13 mohawk which CMake version?
18:13 sivoais the library itself is nice, but typical overcomplicated C++ with templates
18:13 sivoais what I have here is cmake 3.2.2
18:14 mohawk what's the specific problem you're trying to solve at this point?
18:15 sivoais ok, I'll link you the repository. It's not supposed to be up at this point since it has no LICENSE yet
18:17 mohawk i won't tell anyone
18:21 mohawk what's the status of the Changes file? it doesn't look like CPAN::Changes is seeing any content?
18:22 vicash sivoais: if you want CMake to pick up $libname-$version you need to do "target_link_libraries($TARGET $LIBNAME-$VERSION) and it willpick it up
18:22 sivoais vicash: I know, but I'm doing something the other way ;-)
18:22 vicash sivoais: if the $LIBNAME-$VERSION has a symlink to $LIBNAME.so then that's great it will auto-pickup
18:23 sivoais I'm trying to get CMake to give me the flags so that I can use them in my own Makefile
18:23 * mohawk heaves a sigh of relief - he knows nothing about CMake
18:23 vicash sivoais: oh ok. that's easy too
18:24 vicash sivoais: when you run "cmake $SRCDIR" then after that you have to run $(MAKE) which if you run as "make VERBOSE=1" that gets picked up since CMake generates a VERBOSE flag and you can see all the gory details.
18:24 vicash that's how i do it
18:25 vicash i have been using Cmake for all my projects in C/C++ for over 10 years now (since version 1.8 i think) so I can tell you many hacks
18:26 sivoais yeah, I did that with a small temp project and grabbed the paths that way
18:26 vicash if you want CMake to generate Verbose Makefiles all the time, you can set the CMAKE_VERBOSE_MAKEFILE option to 1 or TRUE in CMakeLists.txt to generate verbose makefiles that you can then peruse
18:27 vicash or like this "cmake -DCMAKE_VERBOSE_MAKEFILE=1 $SRCDIR" . but you have to delete the CMakeCache.txt first so that it is regenerated and picks up this option.
18:27 sivoais Don't know if this will explain my error, but this is how my build fails <https://gist.github.com/3e2b6e2ceef21fa7c8c1>
18:27 sivoais I know I'm doing something a bit silly
18:29 vicash sivoais: let's do the dumb checks first. are you sure that the symbol that is being requested is present in the libraries listed ?
18:30 vicash sivoais: you can list symbols in the library using "nm"... and demangle them using c++filt
18:31 vicash another thing I believe is a possible error is that -Ilib should be -Llib since while linking the link directories are linked with -L and not -I which is for header files
18:32 vicash sivoais: nm $LIB | grep -i $SYMBOL | xargs c++filt
18:32 sivoais ok, giving that a go. I just realised that two of the libs didn't make it back into my makefile after debugging
18:33 sivoais now I just need to make sure the other two are in where I think they are
18:33 vicash ok. remember to use -L/usr/local/lib and let gcc do its job... using /usr/local/lib/$LIB is good but not portable
18:35 opkick [pdla] mohawk2 pushed 1 new commit to changesfix: http://git.io/vGZjr
18:35 opkick pdla/changesfix 1f7b832 Ed J: Changes up to date
18:35 sivoais Thanks, I needed to get rid of that. yeah, I grabbed that from the CMake output.
18:35 opkick [pdla] mohawk2 opened pull request #12: Changesfix (master...changesfix) http://git.io/vGZjX
18:37 vicash sivoais: let me know if you need any other help. you can PM me if the iRC logs will get clogged with this info
18:38 mohawk sivoais, could you review #11, #12 and tell me if you want me to merge #7?
18:39 mohawk actually, given #7 failed the CI it might want some love
18:40 vicash sivoais: another thing is you're using static linking. that may fail on various systems due to the fact that they may not have been compiled with -fPIC and some of the code is compiled with -fPIC etc. if you control all the compilation of ITK then i would suggest compiling everything with -fPIC if possible so that linking is smoother with applications that do use fPIC
18:42 opkick [pdla] mohawk2 merged master into sf#394: http://git.io/vGnvq
18:43 sivoais vicash: thanks, I think I'm making some more progress on this. Or at least making the build a little more portable
18:43 opkick [pdla] mohawk2 deleted sf#394 at 4a6383b: http://git.io/vGnvR
18:44 vicash sivoais: on line 2 of your gist why does it have a heart emoji instead of "fault" ?
18:45 opkick [pdla] zmughal commented on issue #11: Looks good! I'm not sure about some of the reasoning in `coretarget()`, but if it all works, I trust you got it right and we should merge. :-)... http://git.io/vGnvb
18:45 opkick [pdla] mohawk2 force-pushed pdldocfix from d484cfd to 0cbcc43: http://git.io/vGnvx
18:45 opkick pdla/pdldocfix 0cbcc43 Ed J: Add PDLA::Doc::add_module to pdlpp_postamble
18:46 sivoais vicash: hahah, I set my shell to show a heart instead of $
18:46 sivoais the branch I'm on is the segmentation branch. Image segmentation, not segfault
18:46 vicash ah ok. makes sense.
18:46 opkick [pdla] mohawk2 commented on issue #11: It only takes effect if you do `make core[test]`. It's partly to keep the code somewhere since I want to bring that functionality into EUMM. http://git.io/vGnfR
18:47 mohawk or so sivoais claims
18:47 opkick [pdla] mohawk2 opened pull request #13: Add PDLA::Doc::add_module to pdlpp_postamble (master...pdldocfix) http://git.io/vGnfK
18:48 vicash sivoais: also you seem to be linking c++ and C code together. since that is a linker command you can just use the C++ compiler instead of "cc std=c99" stuff maybe.
18:50 opkick [pdla] mohawk2 deleted coretarget-parallel at 1821b3c: http://git.io/vGnJW
18:52 opkick [pdla] mohawk2 force-pushed changesfix from 1f7b832 to 5ccb221: http://git.io/vGnUJ
18:52 opkick pdla/changesfix e7d5e4d Ed J: Revert accidental historical revisionism
18:52 opkick pdla/changesfix 5ccb221 Ed J: Changes up to date
18:53 opkick [pdla] mohawk2 merged changesfix into master: http://git.io/vGnUt
18:53 opkick [pdla] mohawk2 closed issue #10: Revert Changes* PDLA->PDL, update http://git.io/vGZyr
18:53 opkick [pdla] zmughal force-pushed test-cleanup-3 from 33bb71e to ef2e045: http://git.io/vGZyt
18:53 opkick pdla/test-cleanup-3 ef2e045 Zakariyya Mughal: test cleanup part 3...
18:53 opkick [pdla] mohawk2 deleted changesfix at 5ccb221: http://git.io/vGnUm
18:54 mohawk sivoais, if you're busy, want me to rebase and fixup tc3?
18:54 mohawk i want to merge and release it before i start on the Big Split
18:57 opkick [pdla] zmughal created ci-fix-readme from master (+0 new commits): http://git.io/vGnT2
18:57 mohawk cor
18:57 opkick [pdla] mohawk2 pushed 1 new commit to master: http://git.io/vGnTr
18:57 opkick pdla/master d05d4ad Ed J: Update Changes
18:58 mohawk sivoais, if i may be so bold as to suggest sequencing, the Big Split is blocked on tc3
19:00 opkick [pdla] zmughal pushed 1 new commit to ci-fix-readme: http://git.io/vGnkm
19:00 opkick pdla/ci-fix-readme 48da09e Zakariyya Mughal: change the links for the badges to point to the new CI locations for pdla
19:00 opkick [pdla] zmughal opened pull request #14: Fixes the README badges for the CI (master...ci-fix-readme) http://git.io/vGnkl
19:01 sivoais mohawk: the only issue was the DynaLoader version (2.013 v. 2.0130001)
19:01 sivoais so I rebased on origin/master and kicked off another CI build
19:03 mohawk sivoais, i fixed that in 57865cb16c58c47cb8cfab461df3bf828c56339d
19:04 sivoais ok, so it should work this time *crosses fingers*
19:04 mohawk grin
19:04 mohawk i'm keen to push through tc3 for the above mentioned reasons
19:04 mohawk if you're on other matters, i can have a go
19:04 mohawk else, i'll bring the alien-hdf4 (current code, not AB) up to date
19:06 opkick [Alien-HDF4] zmughal commented on issue #2: PDLA will need this for post-split improvements of the PDLA-IO-HDF dist. http://git.io/vGnIB
19:09 sivoais mohawk: we'll have to wait a bit on tc3. ETOOMANYBUILDS
19:10 mohawk sivoais, i'd like to take it on, for the reasons mentioned
19:10 mohawk are you comfortable with that?
19:11 sivoais mohawk: go right ahead. But I think it was just the DynaLoader version that was breaking the build
19:11 mohawk oh, great
19:11 mohawk i guess we'll find out soon!
19:11 mohawk exciting times
19:12 mohawk i'm just crossing the alienhdf4 over to PDLA
19:17 opkick [pdla] mohawk2 force-pushed alienhdf4 from e85440e to c20a0e5: http://git.io/vGnqa
19:17 opkick pdla/alienhdf4 c20a0e5 Ed J: Use Alien::HDF4 in PDLA::IO::HDF build
19:17 travis-ci PDLPorters/pdla#13 (master - 1821b3c : Ed J): The build passed.
19:17 travis-ci Change view : https://github.com/PDLPorters/pdla/compare/4a6383be5113...1821b3cd1e78
19:17 travis-ci Build details : https://travis-ci.org/PDLPorters/pdla/builds/77843316
19:18 mohawk facebook's motto was, until 2014, "move fast and break things"
19:18 mohawk one aims to keep the breakage minimal, but it's the fast movement that achieves things
19:18 opkick [pdla] mohawk2 opened pull request #15: Use Alien::HDF4 in PDLA::IO::HDF build (master...alienhdf4) http://git.io/vGnq9
19:20 opkick [pdl] mohawk2 deleted sf#394 at b26afaa: http://git.io/vGnms
19:23 mohawk woops, left the "changesfix" build going, mea culpa
19:26 opkick [pdla] mohawk2 comment on pull request #14 48da09e: Should be /pdla? http://git.io/vGnYb
19:27 opkick [pdla] zmughal commented on pull request #14 48da09e: You're right. Fixing! http://git.io/vGnOs
19:27 mohawk magic
19:27 mohawk i can almost taste quick builds
19:28 opkick [pdla] zmughal force-pushed ci-fix-readme from 48da09e to a3559c7: http://git.io/vGnT2
19:28 opkick pdla/ci-fix-readme a3559c7 Zakariyya Mughal: change the links for the badges to point to the new CI locations for pdla
19:29 mohawk great! i'll merge (and kill the master build)
19:29 opkick [pdla] mohawk2 commented on issue #14: Great work! I'll merge (and kill the master build). http://git.io/vGn3U
19:31 opkick [pdla] mohawk2 force-pushed ci-fix-readme from a3559c7 to 42c6235: http://git.io/vGnT2
19:31 opkick pdla/ci-fix-readme 42c6235 Zakariyya Mughal: change the links for the badges to point to the new CI locations for pdla
19:34 opkick [pdla] mohawk2 deleted ci-fix-readme at 42c6235: http://git.io/vGn3F
19:37 mohawk green lights so far on tc3
19:38 mohawk sivoais, if you could review #13 and #15, those are pre-split ones too
19:38 sivoais not on appveyor
19:38 sivoais the problem is another approx test in lut.t
19:38 mohawk appveyor is less important to me right now
19:39 mohawk yes, there was a travis fail there too
19:39 sivoais ok, let me TODO that lut.t test and then merge
19:39 mohawk feel like TODO-ing all the tapprox there?
19:39 mohawk great
19:39 mohawk and then please also zap the branch etc :-)
19:40 opkick [pdla] zmughal force-pushed test-cleanup-3 from ef2e045 to b5f78a3: http://git.io/vGZyt
19:40 opkick pdla/test-cleanup-3 b5f78a3 Zakariyya Mughal: test cleanup part 3...
19:40 mohawk the way i'm seeing the sequencing is: core stuff that's nice to have we do now or soon
19:41 mohawk but pre-split, i want to get in all the ready non-core stuff
19:41 mohawk actually, test-pgplot qualifies too so i'll attack that now
19:42 sivoais needed to rebase on origin/master
19:42 opkick [pdla] zmughal force-pushed test-cleanup-3 from b5f78a3 to e8f31ef: http://git.io/vGZyt
19:42 opkick pdla/test-cleanup-3 e8f31ef Zakariyya Mughal: test cleanup part 3...
19:44 opkick [pdla] zmughal commented on issue #7: This should be good to go now that I TODO'd the `approx` tests in `lut.t`, right? Want to wait for the CI to be green? http://git.io/vGnZe
19:46 sivoais mohawk: are you going to start over on the pdl-core-only{,-update} branch?
19:46 mohawk i will see how difficult it is, but i'll start from there
19:47 opkick [pdla] mohawk2 commented on issue #7: I don't. Storm in! http://git.io/vGnZF
19:47 opkick [pdla] mohawk2 force-pushed test-pgplot from c0a9975 to efc40ca: http://git.io/vGnZx
19:47 opkick pdla/test-pgplot 9bd5d5a Ed J: Only build PDLA::Graphics::PGPLOT if PGPLOT installed
19:47 opkick pdla/test-pgplot 03ad211 Ed J: Tidy t/pgplot.t, allow window to show if PDLA_INT set
19:47 opkick pdla/test-pgplot efc40ca Ed J: Add t/pgplot.t test with 2 windows
19:48 opkick [pdla] zmughal merged test-cleanup-3 into master: http://git.io/vGnnq
19:49 opkick [pdla] mohawk2 opened pull request #16: Test pgplot better (master...test-pgplot) http://git.io/vGnn3
19:49 mohawk cor
19:49 mohawk sivoais, could you review #16 as well?
19:51 opkick [pdla] mohawk2 deleted test-cleanup-3 at e8f31ef: http://git.io/vGnng
19:52 opkick [pdla] zmughal commented on pull request #16 efc40ca: I like this. Much better. http://git.io/vGnnb
19:52 opkick [pdla] mohawk2 force-pushed test-pgplot from efc40ca to d70dcef: http://git.io/vGnZx
19:52 opkick pdla/test-pgplot 713aecb Ed J: Only build PDLA::Graphics::PGPLOT if PGPLOT installed
19:52 opkick pdla/test-pgplot 384c6ab Ed J: Tidy t/pgplot.t, allow window to show if PDLA_INT set
19:52 opkick pdla/test-pgplot d70dcef Ed J: Add t/pgplot.t test with 2 windows
19:53 mohawk (force push was after rebase to ensure no conflict with tc3)
19:55 mohawk sivoais, #15 passed travis so if you're happy it's ready to merge
19:55 opkick [pdla] zmughal commented on issue #16: Looks good! :ok_hand:... http://git.io/vGncg
19:55 mohawk awww yeah
19:57 opkick [pdla] zmughal commented on issue #15: Could you add `Alien::HDF4` to the `appveyor.yml` as well? http://git.io/vGnCe
20:02 opkick [pdla] mohawk2 commented on issue #15: Instead I've been bold and added it as a `CONFIGURE_REQUIRES` dep. That'll mean wanting to run it through Travis again as well as Appveyor. http://git.io/vGnWc
20:02 mohawk oh dear, the pgplot test is failing
20:04 opkick [pdla] mohawk2 force-pushed alienhdf4 from c20a0e5 to 2edeb0e: http://git.io/vGnqa
20:04 opkick pdla/alienhdf4 2edeb0e Ed J: Use Alien::HDF4 in PDLA::IO::HDF build
20:05 mohawk ...because i am an idiot and hadn't switch the PDL:: to PDLA::
20:05 opkick [pdla] mohawk2 force-pushed test-pgplot from d70dcef to 4999a2f: http://git.io/vGnZx
20:05 opkick pdla/test-pgplot 4999a2f Ed J: Add t/pgplot.t test with 2 windows
20:11 mohawk sivoais, could you on appveyor cancel builds 55-58, and 60 and 61?
20:11 mohawk they're all superceded
20:12 mohawk nice
20:14 mohawk uh, i think i needed 61 after all - any way to restart?
20:14 sivoais I can start a new one
20:14 mohawk ok
20:15 mohawk i'm not overly fussed about appveyor at this point, thanks to the temporary MFBT in effect
20:15 mohawk however, it's definitely a nice to have to reduce breakage control later :-)
20:16 sivoais mfbt ? mtbf?
20:16 mohawk Move Fast...
20:16 sivoais ah... so not mean time between failures  ;-)
20:17 mohawk well, in a manner of speaking maybe it is ;-)
20:17 mohawk i'm desperate to collapse down the number of open branches
20:17 mohawk long-running branches are a burden
20:17 mohawk and we're making great progress
20:17 mohawk all the core ones are low-priority now since that will be better addressed post-split
20:17 mohawk the moment these non-core ones are merged, it's splits ville
20:18 mohawk while we wait for the CI, i'll just think aloud about a plan
20:19 mohawk key point is to have a core-using module, and a core-only module
20:19 mohawk therefore i can use pdla-stats and pdla-core
20:19 mohawk i'll make repos for them both now
20:19 mohawk and i'll ask you to enchant them :-)
20:21 sivoais mohawk: just rename pdl-core :-P
20:22 mohawk good call, i'll delete https://github.com/PDLPorters/pdla-core
20:23 mohawk https://github.com/PDLPorters/pdla-stats
20:23 mohawk needs a kind of magic
20:24 * mohawk sparks up queen greatest hits 2
20:25 sivoais nice
20:25 mohawk sivoais, does pdla-c need enchanting for appveyor as well?
20:25 mohawk (looking at the issues)
20:25 sivoais right, it does
20:25 sivoais that's just me enabling the build
20:25 sivoais should only take a min
20:26 mohawk magic
20:26 sivoais and it is here <https://ci.appveyor.com/project/zmughal/pdla-core>
20:26 mohawk cool
20:26 mohawk appv is low-priority for now :-)
20:27 mohawk so hdf4 looks good on travis, happy with a merge?
20:28 mohawk while i wait for an answer there, i'll make pdla-stats have stuff
20:31 mohawk sivoais, how did you populate pdla?
20:31 sivoais git push --mirror
20:31 mohawk was it by adding both as remotes, then pushing --mirror, or what?
20:31 mohawk ha ha, ok
20:31 sivoais <https://github.com/PDLPorters/pdla/issues/2>
20:32 mohawk magic
20:33 mohawk that info needs to go into devops
20:33 opkick [pdla] zmughal comment on issue #15: Looks good.... http://git.io/vGnEj
20:35 opkick [devops] zmughal pushed 1 new commit to master: http://git.io/vGnuo
20:35 opkick devops/master a242978 Zakariyya Mughal: add a note about how to mirror with git
20:35 mohawk along with: perl -pi -e 's#PDL([^A])#PDLA$1#g' $(gg -l PDL)
20:36 mohawk that git push pattern will be very useful for populating pdla-core
20:37 opkick [devops] zmughal pushed 1 new commit to master: http://git.io/vGnzR
20:37 opkick devops/master 7dd84bd Zakariyya Mughal: add a note about how to rename the namespace from PDL to PDLA
20:38 mohawk nice
20:39 mohawk also needs to mention not applying that change to eg Changes
20:39 mohawk also, a modified change for README.md
20:42 opkick [devops] zmughal pushed 1 new commit to master: http://git.io/vGngK
20:42 opkick devops/master f5359c6 Zakariyya Mughal: add a checklist for renaming a namespace
20:49 mohawk sivoais, i'm feeling twitchy because i did stuff to pdla-stats and no notification here :-)
20:50 opkick [pdla] mohawk2 comment on issue #15: It's an issue but it's probably my fault on some level. I have dmake available here for fast dev when we reach that stage. Merging. http://git.io/vGnaF
20:50 opkick [devops] zmughal opened issue #2: fix Appveyor and GitHub org permissions http://git.io/vGnaN
20:52 opkick [pdla] mohawk2 closed pull request #15: Use Alien::HDF4 in PDLA::IO::HDF build (master...alienhdf4) http://git.io/vGnq9
20:52 opkick [pdla] mohawk2 deleted alienhdf4 at 2edeb0e: http://git.io/vGnVs
20:54 opkick [pdla] mohawk2 force-pushed test-pgplot from 4999a2f to e36076a: http://git.io/vGnZx
20:54 opkick pdla/test-pgplot 697bd40 Ed J: Only build PDLA::Graphics::PGPLOT if PGPLOT installed
20:54 opkick pdla/test-pgplot cdb2cb5 Ed J: Tidy t/pgplot.t, allow window to show if PDLA_INT set
20:54 opkick pdla/test-pgplot e36076a Ed J: Add t/pgplot.t test with 2 windows
20:55 opkick [pdla] mohawk2 closed pull request #16: Test pgplot better (master...test-pgplot) http://git.io/vGnn3
20:55 opkick [pdla] mohawk2 deleted test-pgplot at e36076a: http://git.io/vGnVx
20:56 mohawk cleared for takeoff
21:01 opkick [pdla] mohawk2 pushed 1 new commit to master: http://git.io/vGnrG
21:01 opkick pdla/master d6b9b81 Ed J: Update Changes, 2.013002
21:04 mohawk POSTing upload for PDLA-2.013002.tar.gz to https://pause.perl.org/pause/authenquery
21:04 mohawk hell to the yeah
21:08 * sivoais forgot to eat lunch
21:13 sivoais figured out my bug! I needed an extern "C" bit in what I was linking to
21:18 travis-ci PDLPorters/pdla#33 (master - d6b9b81 : Ed J): The build passed.
21:18 travis-ci Change view : https://github.com/PDLPorters/pdla/compare/e36076a0a3bf...d6b9b8119b5a
21:18 travis-ci Build details : https://travis-ci.org/PDLPorters/pdla/builds/77854188
21:20 mohawk oh wow
21:22 sivoais yeah, the C++ code thought the functions I was calling needed to be mangled.
21:24 mohawk been there a bit, with my experimental "make EUMM work transparently with XS C++"
21:30 opkick [pdla] mohawk2 pushed 1 new commit to master: http://git.io/vGnP7
21:30 opkick pdla/master b3f4d01 Ed J: Update Changes, 2.013003
21:30 opkick [pdla] mohawk2 tagged pdla2.013003 at master: http://git.io/vGnP5
21:32 mohawk and another one, because of versions
21:32 mohawk sigh
21:43 mohawk sivoais, do you have a view on kmx's GENERATED thing for metacpan purposes?
21:43 mohawk mine is broadly that a better way is to have *.pd -> *.pm with Inline::Pdlapp in it
21:44 mohawk (with some Inline::Module magic to make that be correctly installable)
21:46 sivoais i'm sorry, I haven't looked into that yet
21:46 sivoais so I can't give a good answer until I have :-)
21:47 travis-ci PDLPorters/pdla#34 (master - b3f4d01 : Ed J): The build passed.
21:47 travis-ci Change view : https://github.com/PDLPorters/pdla/compare/d6b9b8119b5a...b3f4d01d7869
21:47 travis-ci Build details : https://travis-ci.org/PDLPorters/pdla/builds/77856821
21:47 mohawk grin
21:47 mohawk that is a good answer
21:54 opkick [pdla] mohawk2 created removeGENERATED at d66db9b (+0 new commits): http://git.io/vGnQN
22:12 opkick [pdla] devel-chm commented on issue #3: wow http://git.io/vGnFZ
22:20 chm joined #pdl
22:22 chm mohawk, sivoais: hi, back to a real pc
22:22 chm Trying to catch up.  Seems like a lot is happening since this AM.
22:23 chm mohawk, sivoais: I can't tell from the log but you might consider no-index for the PDLA::*
22:24 mohawk to what end?
22:24 chm I expect once things are split off and cleaned up, you won't need much of the existing cruft.
22:25 chm mohawk: I had to aquire all the perms for PDL modules to make an official release.
22:25 mohawk what cruft would that be?
22:25 chm In some cases, I was even given all the authors modules---many I may not even know.
22:27 chm I would say that 382 module namespaces seems a bit excessive for "throw away names"
22:27 mohawk why do you say "throw away"?
22:27 chm Assuming things get cleaned up, I would expect much of them would move to external module
22:28 chm The others that get replaced by new API/ABI... will live on forever.
22:28 chm mohawk: Thanks for the hv_fetch to hv_fetchs upgrade.
22:29 chm PDL only recently bumped up from 5.8.* to 5.10.* for the perl version.
22:29 mohawk yes, i know
22:29 mohawk i told you i wasn't in favour of that
22:29 mohawk but regardless, here we are
22:30 chm you wanted to stay at 5.8.*?
22:30 mohawk yes
22:30 chm why?
22:30 mohawk if you want to be an important, toolchain-like module, then you have to support perls going back that far
22:31 chm I don't think that the current PDL architecture was _ever_ that implementation.
22:31 mohawk indeed
22:32 chm I think the new PDL3 core approach (number just to indicate the next get stuff) should be the default data blob for perlers needing some type of data/array/compute functionality.
22:32 chm It might even make sense to have something like PDL::Tiny that starts out as pure perl.
22:33 chm We could use Roles to upgrade to more XS-ish capabilities.
22:33 mohawk so you said a year ago
22:36 sivoais to be honest, moving to roles is a big jump. I learned a lot about some approaches to that over the past year
22:36 sivoais but I think the most pressing task now is splitting PDL to ease maintenance
22:37 mohawk not just maintenance
22:37 mohawk development
22:37 sivoais because while PDL3 is being implemented, PDL/PDLA will need to keep going
22:37 mohawk if you think of it as a thing to be afraid of, and maintained, you will do a poor job
22:37 sivoais yep, both parts of the cycle
22:37 chm Role or not, the idea was we could start pure perl as a base.
22:38 chm I'm fine with the split, especially since there seems to be a way forward without breaking PDL-2.x
22:38 mohawk you're fine with the split?
22:38 chm mohawk: Please contribute to the pdl-devel thread so we can get the rest of the devels in on things.
22:38 mohawk i did in june
22:39 mohawk i am not a huge fan of repeating myself
22:39 chm You're not a fan of being clear either.
22:39 mohawk nice little personal dig, chris
22:39 mohawk super factual as well
22:40 chm Just because you said something in one context and can now read it in another and it "makes sense" to you, doesn't mean that will work for others.
22:40 chm you should talk about the personal digs....
22:41 chm I'm just trying to close the loop with other devels without putting words in anyone's mouths.
22:41 chm At the very least, you could include the links to the emails that you keep referring to.
22:41 chm I'd like to see them in the current context as well.
22:46 sivoais mohawk, chm: I think I was the one that posted to the list :-P
22:46 sivoais <http://thread.gmane.org/gmane.comp.lang.perl.pdl.devel/6534/focus=6543>
22:46 mohawk https://sourceforge.net/p/pdl/mailman/pdl-devel/thread/DUB115-DS196882EDB423AD392B7B0E82CD0%40phx.gbl/#msg34140394
22:47 sivoais ok, so we both did :-P
22:48 mohawk i'd like to apologise to everyone here
22:48 mohawk i lied
22:48 mohawk the above email was in may, not june
22:49 mohawk here is another one in may: https://sourceforge.net/p/pdl/mailman/pdl-devel/thread/DUB115-DS198CABD82C89C9DAB9D14182C80%40phx.gbl/#msg34158032
22:49 mohawk here's a request to release a new PDL::IO::GD in early june: https://sourceforge.net/p/pdl/mailman/pdl-devel/thread/DUB115-DS38416181ADA06C5281D77082BF0%40phx.gbl/#msg34186603
22:50 sivoais clearly there's been a communications failure as the threads don't continue as they should
22:51 sivoais if we can't get the communication across, it becomes difficult to work as a distributed team
22:52 chm Thanks for the cross-refs.  I don't recall much discussion in the previous posts.
22:52 mohawk chm, did you in fact receive the emails i have linked to in the archive?
22:52 sivoais and in the cos-design.pdf that I linked earlier, a large part of making an iterative development model work is communication.
22:53 chm I'm not at my email at the moment.  I'll check later.
22:53 chm sivoais: I agree communication is important.  To me it seems that mohawk and my styles don't work together.
22:54 chm Given the work sivoais and mohawk and others are making, it is important to get things set for development.
22:55 mohawk i'd love to hear your plans for that
22:56 chm mohawk: I believe the current PDLA + PDL(stable) approach addresses my concern about PDL stability for non-expert users.
22:57 sivoais so, that's good. What other concerns need to addressed?
22:57 sivoais *to be
22:57 chm I'm just trying to communicate that to the pdl-developers to let them understand what things are changing (PDLA and fast) and what will be stable (PDL which will have PDLA folded in periodically).
22:57 mohawk There is this special biologist word we use for 'stable'. It is 'dead'.
22:57 mohawk -- Jack Cohen
22:58 sivoais right
22:59 chm sivoais: My concern is that one of the new developers doesn't seem to think an accessble PDL that "just works" to new users is important.
23:00 chm mohawk: My experience during my research was that I had to be certain my computations were correct.
23:00 chm in process, I did not upgrade software to maintain provenance of the work.
23:00 sivoais chm: OK. And the PDL (with fast dev on PDLA) approach addresses that, right?
23:01 chm I think users trying to get computation done in perl need that ability.
23:01 mohawk there are better, more certain techniques to manage software versions
23:01 sivoais yes, same. I've been bitten by upgrades that break my data processing piplelines too many times
23:01 mohawk "infrastructure as code"
23:01 chm I'm trying to prevent the same 3 year blight that Numpy had while it was being reworked.
23:02 sivoais chm: do you have a link to that? was that the numpy -> numeric -> numpy change?
23:02 chm With PDL as the underdog, I think having it easy and accessible is important to maintain ming share.
23:02 chm sivoais: yes, that was the one.
23:02 sivoais numarray -> numeric -> numpy
23:03 chm sivoais: I think it was numpy -> numarray -> ... -> NumPy again
23:03 chm sivoais: sorry I didn't really keep too much track of details until things stabilized again.
23:04 sivoais ok, so I think the package stability is addressed by keeping PDL as it is now.
23:04 chm mohawk: I think the need for the PDL "training wheels" will go away as the PDLA/PDL3 moves forward
23:05 chm mohawk: I just think it is critical to PDL to have it "just work"
23:05 mohawk i think you didn't read my last 2 messages
23:05 chm which messages?
23:05 mohawk on here
23:05 mohawk 4 minutes ago
23:05 sivoais "there are better, more certain techniques to manage software versions"
23:06 chm mohawk: better, more certain techniques are great!  I like to see them working before breaking PDL-2.x on the way there.
23:06 mohawk you have literally no idea what i'm talking about
23:07 mohawk and you haven't asked a single question about it
23:07 sivoais mohawk: could you explain what you mean? I think I understand, but only partly
23:07 mohawk vagrant
23:07 sivoais so a VM with the code inside it
23:08 mohawk then provision your shiny new vm with specified versions of perlbrew, perl, cpanm, fixed versions of PDL
23:08 sivoais or carton
23:08 mohawk i haven't tried carton, but that would be the easy way to pin versions
23:08 sivoais run4flat wrote about this wrt to PDL
23:08 mohawk the entire script to execute it would be about 10 lines, and make a nice appendix to each of one's papers
23:09 chm Many users are not able to install VMs on their system.
23:09 mohawk anyone wanting to repro your results could guarantee using exactly the same software
23:09 chm At any rate, my concern is not for expert computer scientist users
23:10 chm The concern was for someone needing to do a matrix computation.
23:10 chm That they be able to "cpanm PDL" and have it just work.
23:10 mohawk anyone who can install PDL using the instructions you had in the README would be capable of doing the above
23:11 chm I think there are more directions to go for ways to handle distribution.
23:11 mohawk whose idea was the use of cpanm, chris?
23:12 chm mohawk: I used that example specifically because you should understand.  Apparently I was wrong.
23:13 mohawk cpanm was one of the easy techniques that i brought along
23:13 chm mohawk: Please let me know (somehow) how we can finish this transition.
23:13 mohawk which transition are you referring to?
23:13 chm All my attempts get argued about, ignored, sniped at, or apparently deliberately misconstrued.
23:14 sivoais I'm not sure where this conversation is going.
23:14 chm I need to get some fresh air, later. o/
23:42 opkick [pdla] mohawk2 created basis (+5 new commits): http://git.io/vGcTh
23:42 opkick pdla/basis d4290e3 Ed J: Rename Lib to Libtmp
23:42 opkick pdla/basis 86cce10 Ed J: Sort MANIFEST* and last chunk of .gitignore
23:42 opkick pdla/basis 752a7d4 Ed J: Move Graphics IO Libtmp Perldl2 + Examples + tests to kitchen-sink
23:44 mohawk i should construct a vagrant box that's got perl 5.22 and PDLA on it
23:46 sivoais <https://github.com/PDLPorters/devops/tree/master/regression-vagrant>
23:46 sivoais in case that helps
23:46 mohawk dang you for stealing all my ideas before i have them
23:46 mohawk for shame, sivoais
23:46 sivoais now you know how I feel about Alan Kay!
23:47 mohawk the swine
23:47 mohawk why precise? trusty's where it's at, man
23:47 * sivoais shakes fist *KAAAAAYY!*
23:48 sivoais precise was LTS, right? or did it get bumped up?
23:48 sivoais ah, trusty is the new LTS
23:48 mohawk precise = 12.04 LTS
23:48 sivoais from <https://wiki.ubuntu.com/LTS>
23:48 mohawk yes, since 16 months ago
23:49 mohawk trusty = 14.04 LTS
23:49 sivoais well... I just went with what I remembered when I first learned Vagrant. I'm a plain ol' Debian user so I can't get the Ubuntu versions straight
23:49 mohawk grin
23:56 opkick [pdla] mohawk2 created basis-coreonly (+1 new commit): http://git.io/vGcIW
23:56 opkick pdla/basis-coreonly 724d27b Ed J: Change NAME to PDLA::Core
23:57 opkick [pdla] mohawk2 deleted pdl-core-only-update at 6e09c6b: http://git.io/vGcI4
23:57 opkick [pdla] mohawk2 deleted pdl-core-only at ef074e3: http://git.io/vGcI0

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