🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
Geth rakudo/main: 16c99ad73a | (Elizabeth Mattijsen)++ | src/core.e/Formatter.pm6
Make sure Format.new uses cached Formatter objects

Otherwise we'd be doing the same over and over again, like the old sprintf implementation.		 09:53
rakudo/main: 8a4f4d8e2d | (Elizabeth Mattijsen)++ | 2 files
RakuAST: =defn first paragraph is the term

So that renderers not need to do anything difficult, especially if the term contains markup codes (which in RakuDoc it can).
Geth rakudo/main: a945fc2364 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/PrecompilationRepository.pm6
Revert "Don't need to use _I version of nqp::radix..."

This reverts commit efa78db6e6aab3062a832465ca6e3809688ac5dc.
Many spectest failures caused by this :-(		 11:09
rakudo/main: fa9cd8ef55 | (Elizabeth Mattijsen)++ | 2 files
Add "rakuast" as key to filter roast on

To allow test-files to be added which currently can only succeed when being compiled with the RakuAST grammar		 11:32
roast: 00d4313d7e | (Elizabeth Mattijsen)++ | 2 files
RakuAST: add tests for q:o// q:format// quote types

  - add file to spectest data
  - mark file as "rakuast" so that it won't affect legacy spectests		 11:45
ugexe I wonder how there were spectest failures... the CI didn't show any -- dev.azure.com/Rakudo/rakudo/_build...324385256b
lizmat these were the tests that failed for me: gist.github.com/lizmat/46ff9e7bea4...25643c9aa4 11:46
and which passed after reverting 11:47
also: after the revert, I've had to nuke my install, because really weird errors related to module loading started occurring 11:48
so I think it messed up my existing install, maybe something you wouldn't see in a CI test on a fresh install ?
ugexe seems like nqp::radix and nqp::radix_I would have to be returning different values for certain strings 12:03
lizmat m: use nqp; dd nqp::radix(16,"9999999999999999999999999999999999999999999999",0,0) 12:07
camelia (-7378697629483820647, 46, 46)
lizmat m: use nqp; dd nqp::radix_I(16,"9999999999999999999999999999999999999999999999",0,0,Int)
camelia (14711957192312533040240131460642968162739895572962580889, 46, 46)
ugexe hmm, well, they both technically match the third element like in the code 12:09
nqp::atpos(nqp::radix(16,$outstr,0,0),2) 12:10
lizmat I think the radix version silently discards overflow 12:13
ugexe `use nqp; for 0..Inf { my $str = "$_"; die $_ if nqp::radix(16, $str, 0, 0)[2] != nqp::radix_I(16, $str, 0, 0, Int)[2] }` hasn't failed for me yet 12:15
lizmat why [2]? the value is in [0] ?
ugexe nqp::atpos(nqp::radix(16,$outstr,0,0),2)
is that not [2]? 12:16
lizmat yeah, but:
out[0] = converted value
out[1] = $number-of-digits-converted
out[2] = offset after consuming digits, -1 if no digits consumed
ugexe if nqp::atpos(nqp::radix(16,$outstr,0,0),2) == 40
that is the code in question
lizmat ah... ok
hmmm
ok, let me re-revert and check the result 12:17
ugexe or maybe try to figure out which value is causing it with like RAKUDO_MODULE_DEBUG or wahtever it is 12:18
lizmat hmmm... initial test now ok 12:26
Geth rakudo/main: 12f681bbc0 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/PrecompilationRepository.pm6
Revert "Revert "Don't need to use _I version of nqp::radix...""

This reverts commit a945fc2364a860d41bf81b1e35f869256ef27f4f.
Apparently my local breakage was caused by something else, and MasterDuke's original commit was not to blame. Sorry for the noise.		 13:12

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!