|
github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm Set by AlexDaniel on 12 June 2018. |
|||
|
00:02
reportable6 left
00:04
reportable6 joined
00:40
frost-lab joined
05:33
shareable6 left,
unicodable6 left,
benchable6 left,
sourceable6 left,
statisfiable6 left,
nativecallable6 left,
reportable6 left,
linkable6 left,
coverable6 left,
evalable6 left,
squashable6 left,
releasable6 left,
quotable6 left,
bloatable6 left,
greppable6 left,
tellable6 left,
notable6 left,
committable6 left,
bisectable6 left,
sourceable6 joined,
nativecallable6 joined,
releasable6 joined
05:34
coverable6 joined,
tellable6 joined,
evalable6 joined,
statisfiable6 joined,
shareable6 joined,
committable6 joined
05:35
benchable6 joined,
greppable6 joined,
squashable6 joined,
notable6 joined,
linkable6 joined,
reportable6 joined
05:36
unicodable6 joined,
bloatable6 joined,
bisectable6 joined,
quotable6 joined
06:02
reportable6 left
06:03
reportable6 joined
07:05
Geth left
07:06
Geth joined
|
|||
| Geth | MoarVM/libzstd-version-check: 099cdd65f8 | (Nicholas Clark)++ | Configure.pl We need at least version 1.0.0 of libzstd, to expose the API we need. Ubuntu 16.04 packages version 0.5.1, so an "exists" check alone for the library is not sufficient - if we jsut do that, we think we can enable our code, but then it fails to compile. |
07:41 | |
| MoarVM: nwc10++ created pull request #1490: We need at least version 1.0.0 of libzstd, to expose the API we need. |
07:42 | ||
|
07:45
domidumont joined
|
|||
| Geth | MoarVM: c7a063e560 | (Stefan Seifert)++ | 2 files Fix corruption of fixkey hash entries Several places in the fixkey hash implementation assumed that entries will be pointers to strings. However the hash can store arbitrary structs. So when those structs contained more than just a pointer, often they overwrote other entries. Fix by replacing all occurences of sizeof(MVMString ***) with control->entry_size |
07:47 | |
| MoarVM: 2d14aa9628 | niner++ (committed using GitHub Web editor) | 2 files Merge pull request #1462 from MoarVM/fix_fixkey_hash_memory_corruption Fix corruption of fixkey hash entries |
|||
| nwc10 | nine: woah, I don't think that that one is correct | ||
| nine | nwc10: that's why I requested you review a month ago :/ | 07:50 | |
| nwc10 | yes, but I got waylaid by everything else | 07:51 | |
| and needed time to figure out what the real problem was | |||
| (sorry) | |||
| nine | Feel free to revert. | 07:52 | |
| Though the old code certainly wasn't correct either | 07:53 | ||
| nwc10 | was there an actual test case that triggered the problem? | ||
| nine | colabti.org/irclogger/irclogger_lo...2021-04-07 | 07:54 | |
| Started the day before: colabti.org/irclogger/irclogger_lo...04-06#l154 | 07:57 | ||
| nwc10 | I think that it might just be that at github.com/MasterDuke17/MoarVM/com...a77cebR269 line 269 (which I am struggling to link to) | 08:02 | |
| that return value should be cast to (MVMString ***) | 08:03 | ||
| and then dereferenced | |||
| but I still have other stuff I need to do first | |||
| But "enabling HASH_DEBUG_ITER causes valgrind to complain" is never a good feeling about how long it's going to take | 08:07 | ||
| Geth | MoarVM: 89043c8407 | (Stefan Seifert)++ | 2 files Revert "Fix corruption of fixkey hash entries" This reverts commit c7a063e560e049daeb50c3c2bb44ebaf17643df6. The author of the original code and CI concur that the change was wrong. Who am I to argue? :) |
08:39 | |
| nine | Wouldn't have happened if the re-run button on Github actually worked :/ | 08:40 | |
| But at least I've found a bug in my build system patch for giving fully detailed test output on failure. That bug caused a build with failing tests to succeed... | 08:41 | ||
| Geth | MoarVM: c8c1b4f2b4 | (Stefan Seifert)++ | src/spesh/args.c Fix spesh missing writes to containers When containers get passed as arguments (i.e. read/write arguments), we collect statistics on these containers' contents and create facts from them. These facts are guarded by the arg guard tree which is used for selecting spesh candidates, so a candidate relying on these facts will only be run, when the facts hold. ... (12 more lines) |
09:00 | |
| MoarVM: c395a48b8a | niner++ (committed using GitHub Web editor) | src/spesh/args.c Merge pull request #1476 from MoarVM/fix_spesh_missing_writes_to_containers Fix spesh missing writes to containers |
|||
| MoarVM: 14d9dd566a | (Stefan Seifert)++ | src/spesh/stats.c Fix missing gc_mark of simstackframe's arg_types Found through: AddressSanitizer: heap-buffer-overflow src/spesh/stats.c:40 in incomplete_type_tuple x62f000000040 is located 960 bytes to the left of 51200- byte region [0x62f000000400,0x62f00000cc00) allocated by thread 0 in setup_bin src/gc/gen2.c:27 |
|||
| MoarVM: 975346d527 | niner++ (committed using GitHub Web editor) | src/spesh/stats.c Merge pull request #1478 from MoarVM/fix_missing_gc_mark_in_spesh_sim Fix missing gc_mark of simstackframe's arg_types |
|||
| MoarVM/master: 4 commits pushed by (Stefan Seifert)++, niner++ | 09:02 | ||
| lizmat | so, is it time for a MoarVM bump ? | 09:44 | |
| nine | CI looks good so far: build.opensuse.org/project/show/ho...rakudo-git | 09:46 | |
| I guess a bump is appropriate :) | |||
| lizmat | bumped | 10:06 | |
|
10:07
linkable6 left,
evalable6 left
10:09
evalable6 joined
10:10
linkable6 joined
|
|||
| Geth | MoarVM/libzstd-version-check: 5ee04f0fac | (Nicholas Clark)++ | Configure.pl We need at least version 1.0.0 of libzstd, to expose the API we need. Ubuntu 16.04 packages version 0.5.1, so an "exists" check alone for the library is not sufficient - if we jsut do that, we think we can enable our code, but then it fails to compile. |
10:38 | |
| nwc10 | AFK again for some time | 10:40 | |
| dogbert17 | nine: is it possible that any of your merges today could/should lead to perf regressions? | 11:22 | |
| commit github.com/MoarVM/MoarVM/commit/c395a48b8a seems to do just that | 11:26 | ||
|
12:02
reportable6 left
12:04
reportable6 joined
12:18
frost-lab left
12:26
frost-lab joined
12:44
frost-lab left
|
|||
| Geth | MoarVM: 5ee04f0fac | (Nicholas Clark)++ | Configure.pl We need at least version 1.0.0 of libzstd, to expose the API we need. Ubuntu 16.04 packages version 0.5.1, so an "exists" check alone for the library is not sufficient - if we jsut do that, we think we can enable our code, but then it fails to compile. |
13:30 | |
| nwc10 | AFK again again | ||
|
13:32
squashable6 left
13:34
squashable6 joined
|
|||
| lizmat | dogbert17 nine looks like HEAD drops test-t from roughly 1.1 seconds to 1.4 seconds :-( | 13:35 | |
| on my machine | |||
| dogbert17 | perhaps nine has a theory as to what's going on | 14:02 | |
| lizmat hopes so | 14:05 | ||
| timotimo | well, one of the commits fixes unsafe optimization for arguments that are containers | 15:35 | |
| could be a part of the performance was relying on an optimization thit could cause trouble | 15:36 | ||
| ugexe | it amazing how fast things can be if they only need to be kinda right most of the time | 15:37 | |
| [Coke] | :) | 16:04 | |
| I remember doing a speed test on something a decade ago and it was much faster than anticipated. turned out it had errored out early and not really done any of the work. | 16:05 | ||
| timotimo | very good | 17:10 | |
| shareholders love this | |||
|
17:15
domidumont left
18:02
reportable6 left
18:03
reportable6 joined
18:29
zakharyas joined
|
|||
| lizmat | reminds me of some people not wanting to wait for a tape backup to finish, so decided to use /dev/null as the tape device :-) | 18:51 | |
| timotimo | reminds me of the client that thought the system can't possibly be doing the work they asked for, so the contractor just put a sleep in there and the client was happy | 19:21 | |
| lizmat | :-) | 19:40 | |
|
19:47
MasterDuke joined
20:14
zakharyas left
20:15
zakharyas joined
20:43
zakharyas left
20:44
zakharyas joined
|
|||
| timotimo | was probably a DailyWTF story or so | 21:00 | |
|
21:32
zakharyas left
22:20
dogbert11 joined
22:23
dogbert17 left
23:43
MasterDuke left
|
|||