| Zoffix | ❗❗❗🌟🌟🌟❇️❇️❇️🌠🌠🌠🎉🎉🎉⚠️⚠️⚠️🚧🚧🚧🌈🌈🌈 | 00:16 | |
| PART 2: Upgrade Information for Changes Due to IO Grant Work: rakudo.org/2017/04/03/part-2-upgrad...rant-work/ | |||
| ❗❗❗🌟🌟🌟❇️❇️❇️🌠🌠🌠🎉🎉🎉⚠️⚠️⚠️🚧🚧🚧🌈🌈🌈 | |||
| MasterDuke | Zoffix++ | 00:21 | |
| AlexDaniel | scary shit going on right there! But that's much better than having another newio branch wasted | 00:27 | |
| Zoffix++ | |||
| hm, so I guess cat is not going to happen now? design.perl6.org/S32/Containers.html#List | 00:35 | ||
| otherwise it may be confused with IO::Cat, no? | 00:36 | ||
| Zoffix | That's Cat, not IO::Cat | ||
| And they have similarities: "lazily treat content of a list of strings as one" vs "lazily treat content of a list of file handles as one" | 00:37 | ||
| m: class IO::Cat {}.gist.put # maybe in this way tho | 00:39 | ||
| camelia | (Cat) | ||
| Zoffix | IO::CatHandle it is! | ||
| mst | that's a filehandle that mostly sleep()s when you try to get it to do anything, right? | 00:43 | |
| Zoffix | heh | 00:44 | |
| Geth | roast: d052285878 | (Zoffix Znet)++ | S32-io/indir.t Actually make use of test's description |
01:39 | |
| roast: 04333b3baa | (Zoffix Znet)++ | S32-io/indir.t [io grant] Test &indir fails with non-existent paths by default |
|||
| roast: 73a54489fe | (Zoffix Znet)++ | S32-io/chdir.t [io grant] Remove two fudged &chdir tests They were fudged 5 years ago. Seems the intent is to check chdir for non-existent path fails and not throws, but one test wouldn't work now even if it were unfudged as a sunk Failure would explode. Testing of this functionality will be subsumed by the new tests, to be added shorly. |
02:28 | ||
| roast: a627e59226 | (Zoffix Znet)++ | S32-io/indir.t Fix incorrect order of arguments to indir test |
03:03 | ||
| Zoffix | Spectest never really feels that slow, until you're in a rush :o | 03:05 | |
| Zoffix enables speed mode | 03:06 | ||
| R is aliased to `for i in `git diff --name-only`; do echo $i; scp $i [email@hidden.address] done;' | 03:07 | ||
| And same for roast files :) And then I just shift-tab to terminal on fast VM and run spectest fast, while still able to edit in Atom :) | |||
| (moments like this make me a bit jealous of people who like to use vi for coding) | 03:08 | ||
| Geth | rakudo/nom: 2483d68941 | (Zoffix Znet)++ | src/core/io_operators.pm [io grant] Fix regression in &chdir's failure mode Yesterday's improvement used `.= chdir` to change the CWD, but that means on failure mode, $*CWD ends up containing a Failure object. Fix by returning Failures as is and assigning to $*CWD only on successes. [1] github.com/rakudo/rakudo/commit/a0...4b18086a12 |
03:10 | |
| roast: 86f79ce194 | (Zoffix Znet)++ | S32-io/chdir.t [io grant] Expand &chdir tests |
03:11 | ||
| rakudo/nom: 5464b82007 | (Zoffix Znet)++ | src/core/io_operators.pm [io grant] Improve &*chdir - fail() not throw() - return $*CWD, just like &chdir does - Coerce argument to IO::Path instead of Str and use its .absolute as new CWD. This prevents accidental use of wrong path to chdir to when given some IO::Path objects, since ther .Str does not take its $!CWD into account |
04:12 | ||
| roast: 430ab897a8 | (Zoffix Znet)++ | S32-io/chdir-process.t [io grant] Test &*chdir |
|||
| rakudo/nom: 4c31903e93 | (Zoffix Znet)++ | t/spectest.data [io grant] Add S32-io/chdir-process.t to list of test files to run |
04:22 | ||
| roast: b41a8ad41a | (Zoffix Znet)++ | S02-types/int-uint.t Unfudge passing int64 tests The fudge line covers more than just the newly-passing test, so duplicate the test in an if/else branch to just fudge it for tests that are not for int64 |
04:48 | ||
| Zoffix | ZOFVM: Files=1231, Tests=133141, 112 wallclock secs (21.36 usr 3.09 sys + 2340.76 cusr 126.00 csys = 2491.21 CPU) | 04:58 | |
| ZOF6VM: Files=1231, Tests=133141, 224 wallclock secs | 05:06 | ||
| Geth | nqp: e5f9b825c3 | (Samantha McVey)++ | 2 files Better Configure.pl errors when finding a MoarVM binary w/o version |
06:29 | |
| nqp: 097acbfe77 | (Samantha McVey)++ | tools/lib/NQP/Configure.pm blah |
|||
| samcv | oops. RE 2nd commit | 06:30 | |
| oh well. i tried to Ctrl + C | |||
| was having a problem using ./Configure.pl in nqp. so now i added new error detection so it will tell you if your MoarVM doesn't supply a version but works and need to checkout tags and recompile | 06:31 | ||
| and made the specific errors display at the top instead of at the bottom so easier to find the most relevant info. | |||
| [Tux] | This is Rakudo version 2017.03-115-g4c31903e9 built on MoarVM version 2017.03-88-g21fc7a22 | 06:50 | |
| csv-ip5xs 3.229 | |||
| test 12.392 | |||
| test-t 4.843 - 5.029 | |||
| csv-parser 13.011 | |||
| dogbert17 | hmm, there are several spectest failures | 09:17 | |
| dogbert17 suspect IOninja++ doing io work | |||
| lizmat | Files=1181, Tests=56159, 204 wallclock secs (12.14 usr 4.81 sys + 1208.52 cusr 116.91 csys = 1342.38 CPU) | 09:37 | |
| yes, indeed | 09:38 | ||
| dogbert17 | nqp: blah :) | 10:05 | |
| camelia | Confused at line 2, near "blah :)" at gen/moar/stage2/NQPHLL.nqp:765 (/home/camelia/rakudo-m-inst-1/share/nqp/lib/NQPHLL.moarvm:panic) from gen/moar/stage2/NQP.nqp:915 (/home/camelia/rakudo-m-inst-1/share/nqp/lib/nqp.moarvm:comp_unit) from gen/mo… |
||
| Geth | rakudo/nom: 5ae454976b | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Streamline generic handling of (<+) (>+) Making it about 15% faster: mostly because of using .AT-KEY, thereby skipping the {} step. |
10:32 | |
| roast: 47581980e1 | (Pawel Murias)++ | S02-lexical-conventions/unicode.t [js] Fudge lack of full unicode support |
11:07 | ||
nqp/master: 5 commits pushed by (Pawel Murias)++
|
12:04 | ||
| rakudo/nom: b64f210c7f | (Elizabeth Mattijsen)++ | 2 files Add internal raw_hash method to Setty/Baggy To allow set operators access to internals, so that we can speed up the common cases by orders of magnitude. |
12:13 | ||
| lizmat | hmmm... isn't (>+) the same as R(<+) ? | 12:30 | |
| m: class A { method succ() { ^42 .pick } }; say (A .. *).head(10) # wonder if this is a bug | 12:46 | ||
| camelia | ((A) 27 28 29 30 31 32 33 34 35) | ||
| lizmat | or a DIHWIDT | ||
| Geth | rakudo/nom: 1277fb5875 | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Simplify (>+) as it is just R(<+) |
12:48 | |
| [Coke] | Zoffix: I ended up not having time to review. I hope you got some feedback on your IO stuff. | 12:49 | |
| jnthn | lizmat: What would the bug be? :) | 12:52 | |
| But yeah, I think the doctor would be like "stop taking that stuff" :P | 12:53 | ||
| Geth | rakudo/nom: 224e40f22f | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Make Setty (<+)/(>+) Setty at least 25x faster - this is for worst case in identical Sets of 26 strings - no need to re-calculate .WHICH, so still faster for expensive .WHICHes - shortcuts for empty Settys and Settys with different number of elements |
13:04 | |
| Zoffix | dogbert17: what sort of failures? There should be no failures on HEAD | 13:06 | |
| lizmat | Zoffix: t/spec/S32-io/chdir-process.t 2-4 | 13:15 | |
| t/spec/S32-io/indir.t 10-21, 26, 28-29, 31, 34, 36, 38, 40, 42-43, 50, 52, 58, 62, 66 | |||
| seeing them since sometime last night, possibly earlier | 13:16 | ||
| dogbert17 | Zoffix: what lizmat said | 13:18 | |
| dogbert17 can retry in a bit | 13:20 | ||
| Zoffix | lizmat: and yet no faiures in t/spec/S32-io/chdir.t? | 13:23 | |
| lizmat checks | |||
| Zoffix | That's really weird. Based on the numbers, looks like something to do with $*TMPDIR, but chdir.t does use it just as indir.t does :/ | 13:24 | |
| lizmat | t/spec/S32-io/chdir.t is ok | ||
| Zoffix | :S | ||
| lizmat | I think symlinks are involved | 13:25 | |
| expected: '/var/folders/7_/ndrzv7y14fj0zxjx3fy0hfjm0000gn/T/ | |||
| got: /private/var/folders/7_/ndrzv7y14fj0zxjx3fy0hfjm0000gn/T | |||
| Zoffix | lizmat: which test is that? one of chdir-proccess.t ones? | 13:26 | |
| lizmat | /var is a symlink to /private/var on MacOS apparently | ||
| yeah | |||
| Zoffix | OK, I'll try making my $*TMPDIR a symlink and see if it explodes | 13:27 | |
| lizmat | so perhaps chdirring to a symlink is borked ? | ||
| (or at least the test is borked :-) | |||
| dogbert17 | All tests successful. | 13:29 | |
| Files=1169, Tests=55610, 423 wallclock secs | |||
| Result: PASS | |||
| that's with 3 cores | |||
| Geth | roast: bc1c3a56c4 | (Zoffix Znet)++ | S32-io/chdir-process.t Fix tests to work right when $*TMPDIR/$*HOME are a symlink |
13:35 | |
| Zoffix | That fixes chdir-process... | ||
| Zoffix looks at indir ones | |||
| Geth | rakudo/nom: 38b341aec8 | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Make Mixy (<+)/(>+) Mixy at least 5x faster - this is for worst case in identical Mixes of 26 strings - no need to re-calculate .WHICH, so still faster for expensive .WHICHes - shortcuts for empty Mixes and Mixes with different number of elements |
13:38 | |
| Zoffix | Can't repro the issue there. | 13:40 | |
| lizmat: would you pastebin output of t/fudgeandrun t/spec/S32-io/indir.t ? I wanna see details of what fails and what was expected. | 13:41 | ||
| lizmat | Zoffix: gist.github.com/lizmat/7b2ee0a9abd...2e47823c93 | 13:45 | |
| Zoffix | Thanks. | 13:46 | |
| Geth | roast: e03ccac0b4 | (Zoffix Znet)++ | packages/Test/Util.pm Make make-temp-[file|dir] fully resolve $*TMPDIR This prevents problems when $*TMPDIR is a symlink and, say, is used someplace to make a path relative to it. If the symlink is not the same level as the actual temp dir, the relativenes will have too few/too many steps in it. |
14:01 | |
| Zoffix | lizmat: that fixes it. | ||
| lizmat | :-) | 14:05 | |
| Geth | rakudo/nom: 928a406a8e | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Make Baggy (<+)/(>+) Baggy at least 6x faster - this is for worst case in identical Bags of 26 strings - no need to re-calculate .WHICH, so still faster for expensive .WHICHes - shortcuts for empty Bags and Bags with different number of elements |
14:06 | |
| timotimo | nice. | ||
| Geth | rakudo/nom: 1ebeeb382c | (Elizabeth Mattijsen)++ | src/core/set_operators.pm Broaden acceptance in (<+)/(>+) - all Quanthashes can be considered Settys for (<+)/(>+) - all Baggys can be considered Mixys for (<+)/(>+) |
14:31 | |
| lizmat | afk for a few hours& | ||
| Geth | roast: 86c5f9c4a4 | (Zoffix Znet)++ | S32-io/path.t [io grant] Delete qp{} tests This file is currently not being run and the 4 qp{} tests test a non-existent feature. We likely won't want that feature ever anyway, as ''.IO is much clearer and is just 1 extra character to type. The file will be repurposed for generic IO::Path tests. |
15:18 | |
| [Coke] | Zoffix++ #cleanups galore. | 15:28 | |
| Geth | roast: 3c4e81b656 | (Zoffix Znet)++ | S32-io/path.t [io grant] Test IO::Path.Str works as advertised |
15:37 | |
| roast: ba3e7be375 | (Zoffix Znet)++ | 2 files [io grant] Merge S32-io/path.t and S32-io/io-path.t |
15:40 | ||
| Zoffix | Oh shit. It's my bday \o/ | 15:46 | |
| Zoffix leaves to bake celebratory chicken wings and play celebratory video games | |||
| [Coke] | Zoffix: happy birthday, youngster. | 15:47 | |
| perlpilot | Zoffix: happy birthday! :) | ||
| Zoffix | Thanks. | ||
| stmuk | m: say "\c[BIRTHDAY CAKE]" | 16:19 | |
| yoleaux2 | 23 Jan 2017 23:32Z <samcv> stmuk: can you syndicate my perl 6 blog o/ pl6net? feed cry.nu/feed.xml | ||
| camelia | 🎂 | ||
| timotimo | happy birthday Zoffix! | 16:53 | |
| Zoffix | Thanks. | 18:53 | |
| Time for some celebratory hacking \o/ | |||
| perlpilot | Zoffix++ you really know how to celebrate! :-) | 19:18 | |
| Zoffix | :) | ||
| samcv | oh i said that a while ago. hello stmuk | 19:25 | |
| Geth | rakudo/nom: cb27bce745 | (Zoffix Znet)++ | 2 files [io grant] Clean up &open and IO::Path.open The dance with $handle and .throw is pointless as .new will never return a :U thing. |
19:26 | |
| samcv | time to put my birthday present in my laptop to get double cores :) | 19:30 | |
| also happy bday Zoffix | 19:31 | ||
| Zoffix | Thanks. | ||
| samcv | be back once i have multiplied :-) | ||
| Zoffix | hbday to you too | ||
| samcv | you too :-) | 19:32 | |
| vendethiel | .oO( I guess we should avoid incrementing Zoffix too often ) |
19:39 | |
| (happy bday, Zoffix++) | 19:40 | ||
| stmuk | samcv: yes maybe the bot was lagged .. your blog was syndicated anyway! | 21:17 | |
| lizmat | stmuk: was it ? I don't seem to see it ? | 21:26 | |
| samcv is taking a lot of time to multiply :-( Hope all is well! | 21:28 | ||
| timotimo | it's vendethiel! | 21:32 | |
| lizmat | and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/04/03/...-advances/ | 22:19 | |
| lizmat gets some sleep | 22:29 | ||
| timotimo | gnite lizmat! | ||
| samcv has now multiplied | 22:39 | ||
| heh and since my TEST_JOBS env var is based on nproc + 1 and so is make, don't even have to touch anything | 22:40 | ||
| timotimo | nice | 22:41 | |
| samcv | 2->4 cores; 2.6->2.7Ghz base; and 3.3->3.7Ghz max freq | 22:45 | |
| nice that's a pretty huge jump in max frequency | |||
| double the cache | |||
| timotimo | nice. | 22:47 | |
| but isn't the max only achievable when you're using just a single core? | 22:48 | ||
| i'm not sure how this "turboboost" stuff works | |||
| samcv | i don't think so | 22:49 | |
| timotimo | excruciating to watch an appveyor build do its thing when you're not yet sure if the config is correct ... | ||
| samcv | heh | 22:50 | |
| timotimo | NMAKE : fatal error U1065: invalid option 'j' | ||
| GGGAAAAAHHHHHhhhhhh | |||
| samcv | i know that feel | ||
| timotimo | it boggles my mind that there are people who develop using windows like ... by choice | 22:51 | |
| Zoffix | lizmat++ good weekly | 22:52 | |
| samcv | making rakudo is not much faster | ||
| but running the tests should hopefully take half the time now | 22:53 | ||
| timotimo | yeah, should | 22:59 | |
| i heard using the harness5 is also noticably faster than the harness6 | |||
| samcv | i never did end up getting a camelia sticker | 23:07 | |
| timotimo | t\nqp/019-file-ops.t (Wstat: 0 Tests: 104 Failed: 0) | 23:08 | |
| Parse errors: Bad plan. You planned 107 tests but ran 104. | |||
| t\qregex/01-qregex.t (Wstat: 0 Tests: 781 Failed: 0) | |||
| TODO passed: 342 | |||
| t\moar/08-indexic.t (Wstat: 0 Tests: 97 Failed: 0) | |||
| TODO passed: 10 | |||
| i wonder what that's all about | |||
| how do i get the tests to be verbose on windows with nmake? | 23:37 | ||
| i believe we have a VERBOSE=1 flag or something? | |||
| samcv | uh | 23:38 | |
| timotimo, you want `nmake -n` | 23:39 | ||
| `nmake -n spectest` | |||
| and then it'll give you the command that it runs, then you can type that in with verbose | |||
| that's what I do | |||
| verified nmake does have that option | |||
| timotimo | no i want the prove to not swallow all the test descriptions | 23:41 | |
| samcv | cool cry.nu/coverage/is now dated and shows the nqp version and moarvm version | 23:47 | |
| and it *should* also tell you if there are failed tests below where the date is | |||
| and it would be nice if the links on the right frame would open on the left | 23:48 | ||
| timotimo | target="_name_of_right_frame" | ||
| samcv | yeah | 23:49 | |
| the boxes were resizeable but i broke it | |||
| timotimo | :S | ||
| samcv | they look resizeable at least. that's half way there | ||
| timotimo | will nmake -n actually give me the output from all those tests? | ||
| samcv | it will tell you the command that runs so you can modify it | ||
| to be verbose | |||
| timotimo | oh | 23:50 | |
| we have test-loud |