|
00:08
AlexDaniel left
00:11
AlexDaniel joined
08:02
AlexDaniel left
09:05
apogee_ntv left
09:06
apogee_ntv joined
|
|||
| Geth | nqp/main: e35f8ba03d | (Elizabeth Mattijsen)++ | tools/templates/MOAR_REVISION Bump MoarVM to get object repossession fix |
10:23 | |
| rakudo/main: f890e01c53 | (Elizabeth Mattijsen)++ | tools/templates/NQP_REVISION Bump MoarVM to get object repossession fix Fixes #6507 |
10:52 | ||
| rakudo/RakuAST-default: f0a2404146 | (Elizabeth Mattijsen)++ | 6 files Make RakUAST default for core and programs Specifying RAKUDO_RAKUAST=1 as before will work. Specifying RAKUDO_RAKUAST=0 or RAKUDO_RAKUAST= will switch back to using the legacy grammar for compilation of programs and modules. This incorporates github.com/rakudo/rakudo/pull/6707 (ugexe++) into a single commit with this change to allow for a smooth transition in CI as well. |
11:33 | ||
| rakudo: lizmat++ created pull request #6710: Make RakUAST default for core and programs |
11:35 | ||
| rakudo/RakuAST-default: 75efc97ca1 | (Elizabeth Mattijsen)++ | t/02-rakudo/m-flag-module-spec.t Fix failing test in Windows CI, at least for now |
12:19 | ||
| rakudo/RakuAST-default: ef77b533f8 | (Elizabeth Mattijsen)++ | 7 files Remove remaining macro tests |
|||
| rakudo/RakuAST-default: 0563e019d9 | (Elizabeth Mattijsen)++ | lib/experimental.rakumod Remove support for 'use experimental :macros' |
|||
| rakudo/main: 5 commits pushed by (Elizabeth Mattijsen)++ | 13:02 | ||
| lizmat | alea iacta est! | 13:03 | |
|
13:34
bisectable6 left,
quotable6 left,
bloatable6 left,
committable6 left,
coverable6 left,
notable6 left,
shareable6 left,
tellable6 left,
unicodable6 left,
evalable6 left,
sourceable6 left,
huggable6 left,
greppable6 left,
releasable6 left,
linkable6 left,
benchable6 left,
nativecallable6 left
13:39
bloatable6 joined,
releasable6 joined,
bisectable6 joined,
benchable6 joined,
linkable6 joined
13:40
committable6 joined,
notable6 joined,
sourceable6 joined
13:41
evalable6 joined,
quotable6 joined,
greppable6 joined,
shareable6 joined,
unicodable6 joined,
nativecallable6 joined
13:42
huggable6 joined,
tellable6 joined,
coverable6 joined
|
|||
| [Coke] | (Still trying to setup a windows build env on the new dual boot so I can at least do ad hoc testing as needed. | 14:00 | |
| s/trying/going/ | |||
| evalable6 | Use of uninitialized value $_ of type Any in string context. Methods .^name, .raku, .gist, or .say can be used to stringify it to something meaningful. in block <unit> at /tmp/RpHyIDdzo5 line 1 |
||
| [Coke] | Maybe not have evalable6 try to run that common idiom as code. :| | ||
| I'm getting a failure on the windows build on my new desktop that I've not seen before: | 14:50 | ||
| gist.github.com/coke/47420c7da9c57...cd77831bee | 14:51 | ||
| Maybe we're not passing the right build flags to zmij? | 14:54 | ||
| ugexe - did you testing include the updated zmij tweak? | 14:55 | ||
| ugexe | github.com/MoarVM/MoarVM/commit/b5...de74cfe320 | 14:57 | |
| i would guess you are building rakudo with cmd.exe or something, and that it'd work if you used x86 command line tools prompt or whatever the hell that thing is called | 14:59 | ||
| releasable6 | Next release in ā4 days and ā3 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft | 15:00 | |
| ugexe | "Why it fails for them and not CI: the guard checks the OS, not the architecture, and interp.c never includes <intrin.h>/<immintrin.h>. _tzcnt_u64 is a 64-bit BMI1 intrinsic that doesn't exist in 32-bit MSVC, so the compiler falls back to an implicit extern declaration and the x86 cdecl leading underscore turns it into __tzcnt_u64 at link time. That extra underscore in the error is the tell ā | 15:01 | |
| they're doing a 32-bit x86 MSVC build. Note _tzcnt_u32 in bigintops.c linked fine in their build, which confirms the compiler recognizes the 32-bit form and only the 64-bit one is missing." | |||
| [Coke] | I'm using developer powershell for VS 18 | 15:05 | |
| and in there, it's setup enough to use MSVC (in a non dev powershell, I get the ming tools that came along with... git, I think) | 15:06 | ||
| ugexe | does it work if you use x64 Native Tools Command Prompt for blah bah blah | 15:08 | |
| there is a x86 Native Tools Command Prompt for blah blah blah as well, and im wondering if one of those works | 15:10 | ||
| [Coke] | there is an x64 command prompt (but not a powershell one, weird.) | 15:11 | |
| one sec. | |||
| ugexe | to be clear though it does seem like a moarvm bug. claude suggests we should have logic similar to this (github.com/MoarVM/MoarVM/blob/a6b3...c#L32-L47) for the function you are referencing | 15:12 | |
| [Coke] | yup, the command prompt environment got past that. | 15:16 | |
| ugexe++ | |||
| works. (now I have to setup my poewrshells to default to utf8) | 15:21 | ||
| ugexe: is your work to fix the MoarVM issue we noted in the AST blin runs in moarvm yet? | 16:47 | ||
| (the recent commit to MoarVM reminded me - not sure if that one *IS* that work) | 16:48 | ||
| ugexe | the most recent commit to moarvm is the one that fixes like ~40 modules. you had included it in some prior ast blin runs but not the most recent one it looks like | 16:52 | |
| [Coke] | no, because it required a brnch in moarvm, nqp, and rkaudo to not impact main, and I was too lazy to keep doing that. | 16:54 | |
| As soon as that's pulled into rakudo & built, will do another blin run. | |||
| will also kill the ast-specific repo | 16:59 | ||
| (of blin results) | 17:00 | ||
| lizmat | m: use NativeCall; my size_t $a .= new # this wasn't a problem in legacy | 17:54 | |
| camelia | ( no output ) | ||
| lizmat | m: say Raku.legacy | ||
| camelia | True | ||
| lizmat | feels like camelia is stuck :-( | ||
| meh, already a 9 days ago | 17:55 | ||
| [Coke] | c: 2c066074e1 e.say | 17:56 | |
| committable6 | [Coke], Ā¦2c06607: Ā«2.718281828459045ā¤Ā» | ||
| lizmat | evalable6: use NativeCall; my size_t $a .= new # this wasn't a problem in legacy | 17:57 | |
| evalable6 | (exit code 1) ===SORRY!=== Can only compile an ā¦ |
||
| lizmat, Full output: gist.github.com/68efa16c442cb802fc...44b154d273 | |||
| [Coke] | c: HEAD use NativeCall; my size_t $a .= new | 17:59 | |
| committable6 | [Coke], gist.github.com/e54d6d76acbe0d781e...a99e0be6e5 | ||
| ugexe | why should that work? | 18:00 | |
| m: use NativeCall; my $a = size_t.new; | |||
| camelia | Cannot instantiate a native type in block <unit> at <tmp> line 1 |
||
| ugexe | would you expect that to work? | ||
| lizmat | ugexe: LibGit2 uses that syntax, and it works in legacy, breaks now | 18:01 | |
| ugexe | did it work? i think i just demonstrated it didnt | 18:02 | |
| lizmat | true, that didn't, but the case with .= did | ||
| m: use NativeCall; my size_t $a .= new; dd $a | |||
| camelia | 0 | ||
| ugexe | either we accepts its wrong or we make native types instantiable which the error explicitly says is not something that can be done | 18:05 | |
| lizmat | it currently already says it is not something that can be done on HEAD | 18:06 | |
| LibGit2 needs a PR and a release, will work on that | |||
| (and fork if necessary) | |||
| ugexe | yes, im saying the fact .= new works on legacy is in conflict with the error message i showed when writing that .= thing in the more straightforward way | 18:07 | |
| lizmat | github.com/CurtTilmes/raku-libgit2/pull/11 | 18:37 | |
| Geth | rakudo/main: 64c2f51fc8 | (Elizabeth Mattijsen)++ | src/Raku/ast/variable-declaration.rakumod RakuAST: improve error trying to instantiate a native a bit |
18:47 | |
| lizmat | looks like MoarVM::Profile needs some TLC, will look at that later today or tomorrow | 19:04 | |
| afk& | |||
| ugexe | hmm, may have to thikn about this more | 19:07 | |
| m: use v6.c; my Int:D $x .= new: 42 | |||
| camelia | You cannot create an instance of this type (Int:D) in block <unit> at <tmp> line 1 |
||
| ugexe | m: use v6.d; my Int:D $x .= new: 42 | ||
| camelia | ( no output ) | ||
| ugexe | that sort of changes my mind and think that the legacy behavior actually isn't inconsistent | 19:08 | |
| i'll take a look at it later | 19:09 | ||
|
19:43
finanalyst joined
21:18
finanalyst left
|
|||
| lizmat | m: my int $a .= new | 22:15 | |
| camelia | ( no output ) | ||
| lizmat | m: int.new | ||
| camelia | Cannot instantiate a native type in block <unit> at <tmp> line 1 |
||
| lizmat | "int.new" is an error in legacy as well, so why would "my int $a .= new" work | 22:16 | |
| as that is just short for "my int $a = int.new" | |||
| ugexe | it changed in 6.d to be on the default | 22:29 | |
| github.com/rakudo/rakudo/commit/38b198c992 | 22:32 | ||
| lizmat | so you'd argue that "int" as a constraint is effectively Int:D ? | 22:34 | |
| m: my int $a; dd $a | |||
| camelia | 0 | ||
| lizmat | m: my Int:D $a; dd $a | ||
| camelia | ===SORRY!=== Error while compiling <tmp> Variable definition of type Int:D needs to be given an initializer at <tmp>:1 ------> my Int:D $a<HERE>; dd $a expecting any of: constraint |
||
| lizmat | anyways, bed time for me :-) | 22:35 | |
| ugexe | it might be that change didn't consider natives at all | 22:37 | |
| my int $a = (my int $).new | 22:38 | ||
| that is basically what legacy does^ | |||
| for a `my int $a .= ne` | |||
| m: class A { has int $.x .= new }; say A.new | 22:40 | ||
| camelia | Cannot instantiate a native type in method at <tmp> line 1 in block <unit> at <tmp> line 1 |
||
| ugexe | hmm, it seems like only `my` declarations in legacy worked that way | 22:41 | |
| i officially change my mind again, the rakuast behavior seems more consistent | |||