brrt good * #moarvm 05:39
timotimo good morning brrt 06:27
samcv morning guys
timotimo hey samcv 06:28
brrt so, i've finally found all commits i want rebased on top of master 07:10
nine brrt: cool! That was probably the hardest part 07:23
brrt yeah, the next part is removing all accidentally added stuff from even-moar-jit 07:27
samcv yay 07:29
nine I wonder how many conflicts I'll get with the jit_nativecall branch after your merge 07:31
brrt not too many i hope, since you're not using the internal jit graph builder structure
fwiw, regarding the pain of building a jit graph by hand
i feel you :-)
the problems, or at least what i think is the problem, is that the 'legacy' jit has nonuniform objects, so you have to specify every damn thing by name 07:32
the expr JIT has a single 'type' of object (a node), and it's an integer, and it's stored in an array that can be modfiied by all the usual array methods 07:33
and… it's just much, much easier to do, even 'by hand'
the uniformity is a big part of why we can have templates, and tiling, and pattern matching, etc 07:34
and a universal tree traverser, etc 07:37
nine I have no idea how long my code will survive in MoarVM, but at least I now have a very good idea how the lego JIT works (it's really that simple :)) and what the motivation for even-moar-jit was 07:38
brrt :-) 07:42
samcv: i will publish my slides, yes, but i'ld like to add a transcript at the same time
or at least
some description that approximates the talk i gave with it
samcv ok :) 07:43
Geth MoarVM/jit-legacy-cleanup: 6 commits pushed by (Bart Wiegmans)++ 08:22
nine brrt: is that all the changes? 08:41
samcv i need to start putting together my resume and looking for a job 09:27
brrt not yet 09:43
nine I think, I'm gonna do the unboxing of arguments in the generated native sub body, rather than in JITed code. That way spesh will be able to strip out unneeded unboxing (same as deconts). 09:49
Also a somewhat funny part: at one point I will want to jit the nativecallinvokejit op itself to enable JITing those deconting and unboxing ops 09:51
JITception
eater is there any docs about what the life cycle of a perl6 script is with MoarVM? 10:39
brrt eater: can you please specify what you mean with lifecycle 11:55
Zoffix huggable: internals course 12:04
huggable Zoffix, Rakudo/NQP Internals Course: github.com/edumentab/rakudo-and-nq...s-workshop
Zoffix eater: ^ perhaps you'll find this slightly-outdated course helpful
eater thanks 12:13
MasterDuke timotimo: do i just need your MoarVM branch for the binary heap analyzer change? or also a branch of the analyzer itself? 12:38
yoleaux 04:37Z <samcv> MasterDuke: thanks for the INTERPOLATE info
Geth MoarVM/jit-legacy-cleanup: 0ceb38a9ea | (Bart Wiegmans)++ | 6 files
Replace explicit JIT function return frame check

This greatly simplifies the JIT compiler driver and jump-out logic. MVM_frame_try_return is now called within the JIT-compiled code.		 15:15
MoarVM/jit-legacy-cleanup: 975bbb09b9 | (Bart Wiegmans)++ | 7 files
Wrap dasm_State in MVMJitComiler structure

This changes function signatures, but is not very exciting otherwise. It prevents spilling the dasm_State structure, and we can use it to collect all compiler state in a single object.
timotimo MasterDuke: analyzer has a branch in my own fork 15:28
Geth MoarVM/jit-legacy-cleanup: 6 commits pushed by (Bart Wiegmans)++ 16:09
timotimo i like seeing progress on this :) 16:31
MasterDuke how does `--tracing` work? 20:13
Geth MoarVM: 5afc7bb673 | pmurias++ | src/gc/debug.h
Avoid evaluating arguments to MVM_ASSERT_NOT_FROMSPACE when it's disabled
 20:17
MoarVM: e71d2a9f5a | (Jimmy Zhuo)++ (committed using GitHub Web editor) | src/gc/debug.h
Merge pull request #658 from pmurias/fix-assert-not-fromspace

Avoid evaluating arguments to MVM_ASSERT_NOT_FROMSPACE
MoarVM: MasterDuke17++ created pull request #667:
Align MVM_TELEMETRY_LOG usage statement		 20:18
MoarVM: 7a980ba828 | MasterDuke17++ (committed using GitHub Web editor) | src/main.c
Align MVM_TELEMETRY_LOG usage statement
 20:19
MoarVM: 56e2219b6f | (Jimmy Zhuo)++ (committed using GitHub Web editor) | src/main.c
Merge pull request #667 from MasterDuke17/patch-2

Align MVM_TELEMETRY_LOG usage statement
MoarVM: MasterDuke17++ created pull request #668:
Align --tracing usage statement		 20:25
MasterDuke i set TRACING = 1 in the Makefile.in and then ran moar with --tracing, but i get the following errors: 20:28
Op 159 at <unknown>:1 (/home/dan/Source/perl6/install/share/perl6/runtime/perl6.moarvm:<dependencies+deserialize>) 20:29
Op 159 at <unknown>:1 (/home/dan/Source/perl6/install/share/perl6/runtime/perl6.moarvm:<entry>)
though my perl6 script does run
timotimo i don't think anybody has used moarvm's tracing support in a loooooong time 21:58
look it doesn't even get the op's name for you
MasterDuke timotimo: gonna be afk for a bit, but do you have any thoughts on my INTERPOLATE questions in #perl6-dev? 22:01
timotimo why it gets called so often? 22:03
i'd say "check out the --target=mast" or something, but i'm not sure
AlexDaniel ? Community Bug SQUASHathon blogs.perl.org/users/zoffix_znet/20...athon.html 23:27
lizmat And another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/08/28/...rving-cro/ 23:37

Powered by Cro and the Raku® Programming Language.

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

Thank you!