| [Tux] | This is Rakudo version 2016.06-89-ge9019e3 built on MoarVM version 2016.06-9-g8fc21d5 | 06:39 | |
| test 15.974 | |||
| test-t 9.388 | |||
| csv-parser 16.863 | |||
| jnthn | Ulti: Good to hear! :-) | 08:48 | |
| dalek | p/optimize-with-hoopl: 585423e | (Pawel Murias)++ | src/vm/js/dataflow-optimizer/QAST.hs: [js] Add missing file. |
09:54 | |
| p/optimize-with-hoopl: 1b19c44 | (Pawel Murias)++ | src/vm/js/DataFlowOptimizer.nqp: [js] Send more info about Vars to the -Ofun optimizer |
|||
| p/optimize-with-hoopl: 0e01a21 | (Pawel Murias)++ | src/vm/js/dataflow-optimizer/ (2 files): [js] Define a constant folding fact and transfer function. |
|||
| p/optimize-with-hoopl: 552f787 | (Pawel Murias)++ | src/vm/js/dataflow-optimizer/optimize.hs: WIP |
|||
| nqp/optimize-with-hoopl: bebf60a | (Pawel Murias)++ | src/vm/js/dataflow-optimizer/optimize.hs: | |||
| nqp/optimize-with-hoopl: [js] Turn an add_i with known arguments into a constant. | |||
| tbrowder | A question on handling pod in Pod.nqp: how should one return if a chunk of pod fails a rule? It looks like the grammar for a pod table accepts more than is strictly allowed in S26 so I would like to reject the table with an appropriate message. | 12:15 | |
| dalek | kudo/nom: 516b997 | lizmat++ | src/core/REPL.pm: Simplify REPL error reporting It occurred to me that using the standard error reporting in the REPL may be a bit overkill: 12f12a5 | lizmat++ | src/core/ (2 files): Fix error reporting more generally This reverts 516b997a4a26 . Basically, <unknown file> will no longer be shown in *any* error reporting. This covers the case that happens in the REPL. So the output is now: > $number = 5; ===SORRY!=== Error while compiling Variable '$number' is not declared ------> <BOL>ā$number = 5; |
12:35 | |
| lizmat | oops, I killed dalek with a single commit ? | 12:36 | |
| github.com/rakudo/rakudo/commit/516b997a4a26d | |||
| jnthn | lizmat: I think retaining the ==SORRY== or so is important too as it is how you know if it's a compile time or runtime error | 12:43 | |
| Completely agree that the unknown file thing is totally useless clutter though. | |||
| It's hugely tempting to get rid of it by regexing it out or so :P | 12:44 | ||
| lizmat | yeah... looking at that now | ||
| jnthn | oh or | 12:45 | |
| Find the place where we append the "in file..." thingy | |||
| And if it's an unknown file just don't emit it at all | |||
| Since it's always useless info | |||
| kudo/nom: 8a2ffac | lizmat++ | src/core/Exception.pm: Show line number if != 1, Juerd++ So the output in that case is now: > my $a = ( * 42,$b) ===SORRY!=== Error while compiling Variable '$b' is not declared at line 2 ------> 42,ā$b) |
13:12 | ||
| kudo/nom: 23d75ed | lizmat++ | src/core/Exception.pm: Final bit of sorry heading bike shedding Felt that "Error while compiling" was missing a colon in the <unknown file> case. |
13:19 | ||
| kudo/nom: 1d4c110 | lizmat++ | src/core/array_slice.pm: Optimize IndicesReificationTarget.push a bit |
14:44 | ||
| [Coke] waves | 15:26 | ||
| RT: 1339; CONC: 7; GLR: 6; JVM: 70; LHF: 1; LTA: 73; NYI: 28; OSX: 6; PERF: 16; POD: 3; PRECOMP: 5; RFC: 22; SEGV: 21; STAR: 1; TESTNEEDED: 28; TODO: 9; UNI: 5; WEIRD: 3 | 15:49 | ||
| japhb | [Coke]++ # Bug tag herding | 16:19 | |
| dalek | kudo/nom: 2a84504 | lizmat++ | src/core/array_slice.pm: Streamline the remaining Seq |
19:01 | |
| sortiz | \o #perl6-dev | 20:07 | |
| lizmat | sortiz o/ | 20:08 | |
| awwaiid | lizmat: In case you didn't notice, I have some other REPL fixes in github.com/rakudo/rakudo/pull/799, including catching errors upon the PRINT part of REPL | ||
| Though interestingly when I do a print-based error on -e I sometimes see "SORRY", whereas here I don't? kinda weird | 20:09 | ||
| dalek | kudo/nom: 5dad919 | (Brock Wilcox)++ | src/ (2 files): Fix behavior on empty lines and empty strings and print-exceptions |
||
| kudo/nom: ab85ab5 | (Brock Wilcox)++ | t/02-rakudo/repl.t: Update REPL test to match current features |
|||
| kudo/nom: e0bcafd | (Brock Wilcox)++ | src/core/REPL.pm: Fix indent |
|||
| kudo/nom: 20ba0ec | (Brock Wilcox)++ | / (16 files): Merge remote-tracking branch 'upstream/nom' into repl-whitespace-fixes |
|||
| kudo/nom: bd0fd78 | lizmat++ | / (3 files): Merge pull request #799 from awwaiid/repl-whitespace-fixes Make List.STORE about 2x faster Which means "my ($a,$b,$c) = (1,2,3)" is now 2x faster |
|||
| awwaiid | woo! | ||
| timotimo | \o/ | ||
| awwaiid hops up and down | |||
| lizmat | indeed, I had missed that | ||
| awwaiid | I'm getting closer to introducing REPL::here; which will be like binding.pry | 20:10 | |
| I feel kinda funny about exporting the top-level REPL name on the one hand, but I also want this to be extremely accessible | 20:11 | ||
| that and my continued quest to remove nqp:: from REPL.pm | 20:13 | ||
| lizmat | ++awwaiid :-) | ||
| sortiz | awwaiid++ | 20:15 | |
| timotimo | YES! | ||
| fantastic | |||
| dalek | kudo/nom: 03bd34e | lizmat++ | src/core/List.pm: Streamline List.Slip Not sure where this will help, but I'm pretty sure it will. |
20:35 | |
| Zoffix | Wow! lizmat++ | 20:38 | |
| vendethiel- | lizmat++ :D | 20:45 | |
| those sound amazing | |||
| dalek | kudo/nom: 5bd5a54 | lizmat++ | src/core/List.pm: Squeeze a few percent out of List.iterator |
21:03 | |
| kudo/nom: ff83625 | lizmat++ | src/core/List.pm: Make the List constructore more memory friendly The difference in CPU usage is within noise levels. However, it does not allocate an IterationBuffer if there are no Slips in the parameter list (which is usually the case when creating a List, I would argue). This could get significantly faster if we had nqp::listhastype(list,Type) op that would return 1 if the list had any element of the given type. |
22:26 | ||
| lizmat | and that concludes the damage I've done today | 22:27 | |
| good night, #perl6-dev! | |||
| timotimo | gnite liz :) | 22:30 |