|
Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes. Set by lizmat on 24 May 2021. |
|||
|
00:02
reportable6 left
00:03
reportable6 joined
00:40
squashable6 left
01:40
evalable6 left,
linkable6 left
01:41
linkable6 joined
01:42
evalable6 joined
02:42
squashable6 joined
03:30
squashable6 left
03:32
squashable6 joined
06:02
reportable6 left
06:03
reportable6 joined
06:38
ilogger2 left
06:39
ilogger2 joined
06:40
patrickb joined
07:00
patrickb left
07:33
frost joined
|
|||
| nine | goo moanin #moavm | 07:39 | |
| Nicholas | good *, nine | ||
| nine | So, I managed to get a spesh dump of the broken frame and it's inlinees by setting a watch point on the spesh slot holding the Str and then reverse-coninuing and calling MVM_spesh_dump manually. | 07:52 | |
| And there may be someting: in the inlined graph of add_object_if_no_sc, there's sp_guardconc r1(2), r1(1), sslot(4), litui32(7) | |||
| This gets changed during inlining to: sp_guardconc r38(1), r29(1), sslot(20), litui32(7) | 07:53 | ||
| Notice the litui32(7) that's unchanged? That's the deopt_idx. And I think it should get updated but doesn't. | |||
| That sp_guardconc doesn't seem to have an INS Deopt Inline annotation which would hold the updated deopt_idx during inlining. | 07:55 | ||
| The instruction before that is sp_guardsf r8(2), sslot(3), litui32(6) and has an annotation | 07:59 | ||
|
08:04
linkable6 left
|
|||
| MasterDuke | github.com/MoarVM/MoarVM/blob/new-...#L527-L532 looks like it should be doing that | 08:24 | |
| nine | But that's only called when there is a Deopt Inline annotation, which this instruction lacks. | 08:25 | |
| MasterDuke | github.com/MoarVM/MoarVM/blob/new-...#L603-L608 it's called for all the deopt kinds, right? not just inline? | 08:29 | |
| nine | Yes, but the instruction doesn't have any annotations at all | 08:35 | |
| MasterDuke | oh, ha | ||
|
09:04
evalable6 left
09:05
evalable6 joined
09:07
linkable6 joined
|
|||
| dogbert17 | nine: so are we now waiting for some words of wisdom from a suitably coffee infused jnthnwrthngtn? | 09:53 | |
| Nicholas | I hope that he didn't have a coffee bootstrapping problem. | 09:55 | |
| nine | dogbert17: either that, or for me to find some more time :) Which...may be a while. There's $dayjob now and some teaching of flight students in theoretical topics this evening. | ||
|
10:47
sena_kun joined
|
|||
| lizmat | jnthnwrthngtn is doing some red tape stuff today, at the other end of Prague, is my understanding | 10:55 | |
| jnthnwrthngtn | I'm just back from it. | 10:56 | |
| Nicholas | \o/ | ||
| jnthnwrthngtn | And grumpy. | ||
| Nicholas | oh. premature \o/ :-( | ||
| jnthnwrthngtn | They got me there to tell me "oh, we don't issue this kind of document any more becuase of Brexit, can you come back tomorrow to do biometric data for the new thing we'll give you?" | 10:57 | |
| I'm like...how boring do you think my life is? No way. | |||
| Anyway, it's now registered in their computer that I satisfied the requirement to visit them to "collect my permit" within 30 days of being told to. | 10:58 | ||
| Nicholas | so the left had knows this, but the right hand that wrote to you to go visit them didn't? | ||
| jnthnwrthngtn | Pretty much that. | ||
| Nicholas | that's stupid | ||
| jnthnwrthngtn | It's very stupid, 'cus it costs half a day to go there/back | ||
| Anyway, they'll see me sometime in autumn, when I can be arsed, because I know that by law I've a year to do the other step. | 10:59 | ||
| Nicholas | that's sort of like the Austrian beaurocracy failure mode, where they are very accurate at following their checklist on their desk, but it isn't always the exact same checklist they put online that you read before you visited. Or there are corner cases. Or whathever. So you don't have all the right ducks lined up in a row. At least, not until visit two. | ||
| jnthnwrthngtn | Yes, last time I went there and took a document that they didn't say they needed "just in case" and then they did. | 11:00 | |
| But given I assume "give biometrics" and "collect card with biometrics" will be 2 more visits, then this whole process will be 4 trips there. | 11:02 | ||
| Although only 2 of them really were for PR and the 2 other two are Brexit fallout. :/ | 11:03 | ||
|
12:02
reportable6 left
13:05
reportable6 joined
14:19
evalable6 left,
linkable6 left,
evalable6 joined
14:35
frost left
|
|||
| [Coke] | PR? | 14:36 | |
| nevermind. | 15:03 | ||
| jnthnwrthngtn | hah, yes, in this context it is not pull request :) | 15:07 | |
| (Permanent Residence, basically "permission to keep living here for a long time") | 15:09 | ||
| [Coke] | best of luck! | 15:10 | |
| (Mainly with navigating bureaucracy!) | 15:11 | ||
| jnthnwrthngtn | Thanks. Well, they granted it, so the decision part is fine, it's just getting a document out of them. :) | 15:15 | |
| [Coke] | nativecall windows failures - tried running one test that has failed on concurrent runs in the debugger while doing a run with concurrency - couldn't get the debug version to fail. | 15:17 | |
|
15:20
linkable6 joined
|
|||
| tbrowder | anyone heard of c static analysis tool frama-c (frama-c.com)? might be worth a look-see for MoarVM | 16:25 | |
|
16:50
linkable6 left,
evalable6 left
17:16
sena_kun left
17:52
linkable6 joined
17:53
evalable6 joined
18:02
reportable6 left
18:05
reportable6 joined
18:20
Altai-man left,
Altai-man joined
21:32
discord-raku-bot left,
discord-raku-bot joined
|
|||
| timo | i am now using my manually assembled split ergo keyboard but i havent learned symbols and such so no programming just yet | 22:33 | |
| japhb | Manually assembled? | 22:34 | |
| timo | aye sir, wiH my own hands | ||
| the layout i stole from a frienq does interesting things i haev not adjusted to, like pressing comma once gives me a comma and a space actually | 22:35 | ||
| japhb | Intersting. | ||
| I've been using a Goldtouch split keyboard for a few years, but before that I had a programmable keyboard, so I could build macros on the fly directly on the keyboard itself. Very useful for certain tasks. | 22:36 | ||
| Kinda wish I could have a programmable Goldtouch though, would be a nice combo. | 22:37 | ||
| timo | i ordered the thing as a kit of two pcbs, a bunch of mechanical keyswitches, keycaps, a bunch of dqiodes and four resistors, two trrs dacks two reset buttons two microcontrollers four slices of acrylic and two smaller ones, screws and brass spacers, two small oled displays | ||
| thats about it | 22:38 | ||
| a friend is sending me leftover rgb leds so i can have undergrond lighting to show layers and other internal state variables | |||
| japhb | Heh. Sounds very techno | 22:39 | |
| Ntts Nttss Nttts | |||
| timo | only if i strobe them haha | ||
| it does not by default have anything to add a speaker to, but one of the uc boards has some pins left over | 22:40 | ||
| japhb | :-) | ||
| timo | and many pins are also broken out at the edges for modding purposes, such as the i2c bus used for the oled and the whatever it is for the rgb leds | 22:41 | |
| japhb | .oO( "There are unused pins! I must connect them to something!" ) |
||
| dogbert17 | timo: does your new keyboard have cat protection :) | 22:44 | |
| timo | depends on what kind you mean | ||
| not really anything to prevent them from stepping on the keys | 22:45 | ||
| dogbert17 | so they can help you code :) | ||
| timo | truly | ||
| dogbert17 | have they shown any interest so far? | ||
| timo | sadly i hadnt realized that the microceontrollers included in this were so incredibly underpowered lol | 22:46 | |
| not terribly much. one already put her leg across it while cleaning herself | |||
| dogbert17 | Cherry switches? | 22:48 | |
| timo | anyway, the controllers are working with 2.5 kilobytes of ram and i think around 26 kib flash | 22:49 | |
| these are gchoc low profile switches by kailh with linear actuation and rather low activation force | |||
| dogbert17 | a bit like cherry red then perhaps | 22:50 | |
| timo | red comes close yes | ||
| cdn.discordapp.com/attachments/348...232516.png | 22:51 | ||
| dogbert17 | interesting design | 22:52 | |
| otaku keycaps it seems | 22:53 | ||
| timo | theres a thumb cluster that.s mostly an arc across the bottom, but also two keys above the innermost two ones that i cant really hit without liftign my hand or turning it around | ||
| splitkb.com/collections/switches-a...nt-keycaps | 22:54 | ||
| also has a picture of the whole thing | |||
| dogbert17 | very cool | 22:55 | |
| timo | it runs qmk which has an incredible feature set to pick from | 22:58 | |
| theres an atmega32u4 in it on both halves here | 22:59 | ||
| [Coke] | (programmable keyboard) I picked up one of keyboardio's on their most recent kickstarter. | 23:19 | |
| (also split) | |||
| timo | those are ceertainly shiny | 23:38 | |
| i was lusting after a moonlander 1 but thats also a rather expensive piece of tech, and theres also shipping and importing fees for that, whereas the kyria gets shipped from the netherlands | 23:39 | ||
