github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
00:04 Geth left 01:31 lucasb left 05:37 MasterDuke left 06:10 domidumont joined
nwc10 ibm-- # needs cookies to render plain pages. Mmm, cookies. 06:41
mmm, I wonder if that is GDPR compliant. 06:42
moritz only if that cookie doesn't store any PII 06:43
nwc10 I guess they condier it "strictly necessary" bceause they've (mis)constructed it to *be* strictly necessary
nine I bet this isn't malice but a side effect of using some arcane Enterprise Java Enterprise Content Management System, Enterprise edition 06:47
Developed by people who have never understood the web.
nwc10 I suspect that it isn't malice either, but all the cool tech seems to do it too
let's do it client side, in a single threaded language.
oh, wait, that was the "and it needs JS" part that I didn't mention 06:48
but the JS is checking the cookie, not the server.
nine of course it does, not worth mentioning
nwc10 aha no, if I run the page in a full-on pr0n window with everything enabled, they are trying to show me a GDPR consent popup. 06:57
Because clearly you *need* all this tracking stuff to render static documentation. 06:58
en.wikipedia.org/wiki/Ping_of_death -- In 2013, an IPv6 version of the ping of death vulnerability was discovered in Microsoft Windows. -- sigh, those who do not learn from history are doomed to repeat it 07:12
sadly I can't find the story of the how the original Xbox was cracked, as this was a similar "not knowing your history" problem. 07:15
Welcome to the future - people are complaining about 720p - old.reddit.com/r/spacex/comments/m...es_thread/ :-) 08:04
dogbert11 nwc10: www.youtube.com/watch?v=XuOm-RTQxVU 08:05
nine I've long wondered about that :)
nwc10 dogbert11: thanks, will watch soon. But there was a lovely written page with about 17 flaws they tried exploiting, and I can't find that currently. 08:06
I watched this one and I was amused: www.youtube.com/watch?v=4BlpONgj74A 08:07
useless bots
"How a pair of Tweezers defeated security on the Nintendo Wii" 08:08
08:14 zakharyas joined 08:30 sena_kun joined 08:56 Geth joined
jnthn morning o/ 09:25
09:30 Geth left
lizmat \o/ 09:32
nwc10 \o 09:33
10:04 LizBot left, LizBot joined
jnthn This GC bug is taking some finding... It's not fromspace at the point we construct the dispatch program, and I see us write barrier it too 10:13
nwc10 rr? 10:15
tea?
lunch?
jnthn Aha! Added in checking if the static frame we use as the root is really the one that owns the inline cache entry. Guess what? It's wrong. 10:21
So we're doing the write barrier...on the wrong root.
nwc10 oh my. Recent commit at fault? or old? 10:22
and is it a pattern? (if "might be", I'll try to have a look for more, once you've pushed it, *and* I've got time)
jnthn Well, it's about the inline cache, so recent as in "part of new-disp"
May be much more recent though; I wonder if I could have confused the applicable static frame when I put in the "bind failure becomes resume" mechanism a while back 10:23
It ain't, apparently 10:28
I put the did the same check in at the entry to recording a dispatch program and it triggers on the typical code path for recording dispatches 10:29
Ah bollocks, looks like copy pasta from months ago 10:30
nwc10 we're having pasta for lunch, but fortunately not recycled 10:31
jnthn Yup, that was it 10:32
10:34 unicodable6 joined
nwc10 \o/ 10:35
timotimo always impressed to see how low-powered these consoles all were 10:37
in terms of clock frequencies of the CPUs, though of course comparing powerpc with x86 isn't 1:1, but also the amount of RAM they had was so low 10:42
nwc10 yes 10:43
jnthn No geth, but the fix is pushed, along with some GC debug checks to catch any similar accidents in the future 10:46
And `make spectest` still looking good with it enabled. 10:47
So I guess that means I've fixed all the known broken and after lunch it really will be time to break more things :) 10:48
lizmat
.oO( that's very raku of you )
timotimo the wii homebrew channel song that plays when you open the "channel" on the UI is really catchy 10:49
www.youtube.com/watch?v=zLDlWNDmR8k this full version has an additional 20 seconds of intro / buildup before the point you'd get when you turned it on on the console 10:50
10:51 Kaiepi left 10:52 Kaiepi joined 10:55 demostanis[m] joined 10:56 Kaiepi left
timotimo i wonder why rakudo doesn't play catchy music when you start it 10:56
lizmat [12:51:00] <lizmat>not sure, you should probably ask that on #moarvm :-) 10:57
[12:55:23] <demostanis[m]>I still can't post in there with matrix
the question was:
[12:49:44] <demostanis[m]>Oh ok, is there anything on the MoarVM website which says which key the binary is signed with? 10:58
jnthn sena_kun ^^ just now on #raku 11:01
sena_kun lizmat, yup, saw that, was not sure what to answer. I think it's a lack on the website code. It does not even provide a button to download the .asc, even though we properly upload it. 11:02
I also wonder what the request is about, to add something like "This RSA key was used to sign packages"? But if it's compromised it means such a label can be replaced as well. 11:05
lizmat well, /me doesn't know either 11:06
lizmat is just the messenger here :-)
11:09 lizmat is now known as lismat, lismat is now known as lizmat 11:15 LizBot left, LizBot joined 11:21 zakharyas left
[Coke] (tracking cookies) this is something I'm fighting with at work, between technology and lawyers. 11:47
... and businessfolk with requirements. 11:48
nwc10 you make "requirements" sound like it's a 4 letter word. :-/ 11:53
[Coke] the business requirements are fine, it's just trying to satisfy all 3 areas. 12:00
nwc10 jnthn: fun, excitement, ASAN - paste.scsys.co.uk/595174 12:04
this wasn't the only one faliilng with all the nodelay
use-after-free 12:05
12:19 dogbert17 joined 12:21 dogbert11 left 12:23 dogbert11 joined 12:25 dogbert17 left 12:59 patrickb joined
jnthn nwc10: hmmm, since that turns out to be on the "legacy" args handling path, I might put that one off to see if the code in question goes away. That said, I'm not really sure how it's happening... 13:39
Uff, how on earth can giving KnowHOW a find_method method cause breakage... 14:32
grmbl, I see how 14:35
nwc10 because it knows that it is Friday afternoon? 14:38
jnthn $ RAKUDO_NEW_DISP=1 ./rakudo-m -e '42.say' 14:46
42
Yay. Setting the env var now rewrites callmethod into a dispatch op. 14:47
So that `42.say` will go via the raku-meth-call dispatcher
I suspect this plan of attack will help to figure out some things, but since the way proto subs get compiled is changing, at some point I'll have to bite the bullet and compile even CORE.setting with new-disp too 15:19
*with RAKUDO_NEW_DISP 15:20
Ugh, a bit sleepy. Think that's all I'll manage today. 15:30
sena_kun jnthn++ # great job 15:33
15:33 zakharyas joined 15:45 Merfont joined
nwc10 jnthn: use more 'coffee'; # I am sure that this is wise ;-) 15:58
nwc10 is actually thinking about beer.
it's a nice sunny evening here.
and $other has calmed me down a bit, so after 3 dry days I'm stopping.
[Coke] Bought cans of spiked cold brew this week. Not sure my body knows what do to with those. 16:15
16:25 Merfont left 16:27 Kaiepi joined
nwc10 "tactical chunder"? :-) 16:33
on the subject of pavement pizza 16:37
jnthn: note to future jnthn - the backtrace changed: paste.scsys.co.uk/595175
jnthn Uff, that's...not nicer
nwc10 aye. 16:38
I'm going to rebuild with valgrind and see if it spots somethign earlier
jnthn Though also suggests we mighta got "lucky" before that and had a valid pointer
The wild pointer one suggests it can be utter junk
nwc10 aye, which makes me hope that valgrind finds something else
jnthn has already moved on to the lighter task of working on one of this month's Comma features 16:39
nwc10 I moved onto beer sitting outside. I slack. I'm sorr. 16:40
jnthn That is my plan in about 15 minutes time :) 16:41
nwc10 I suspect you've earned it more than me. 16:42
jnthn Guess I can only relax for so long, I'm meant to be cooking dal for dinner. 16:44
Although the temptation to just have something delivered is non-zero... :)
nwc10 I had suggested this but it was sort-of-vetoed 16:45
(I am not usually in charge of cooking. But I was considering curry. From the "Italian")
[Coke] I have noted with $spouse that we seem to eat out a LOT more during these trying times. 16:46
(takeout out, not sit down out)
16:50 domidumont left
jnthn We pretty much replaced the number of times we did sit down out with doing takeout out. 16:52
lizmat the takeout out has more or less remained the same for us, the sit down out is now sit down at home with home cooked meals 17:05
nwc10 sorry for bad news, but valgrind doesn't find anything. IIRC one of the poosiblities was that valgrind disabled ASLR, and another was that it ended up with addresses being in the bottom 32 bits 17:06
I shall have a go with rr, but not right now
enjoy whatever food you settle on 17:07
jnthn :) 17:08
Yeah, I've done the feature I was meant to, so now it's time to rest :)
timotimo debugger goes rrrrrrrr 17:18
18:44 vrurg left 18:51 vrurg joined 18:57 vrurg left 20:41 Kaiepi left 20:47 patrickb left 20:51 Kaiepi joined 21:13 zakharyas left 21:28 vrurg joined 21:30 MasterDuke joined 21:33 vrurg left 21:47 Kaiepi left 21:50 vrurg joined 21:52 Kaiepi joined 22:02 Kaiepi left, Kaiepi joined 22:13 dogbert17 joined, dogbert12 joined 22:16 dogbert11 left 22:18 dogbert17 left 22:20 Kaiepi left, Kaiepi joined 22:48 dogbert17 joined 22:51 dogbert12 left 22:52 dogbert17 left, dogbert17 joined 22:59 dogbert11 joined 23:00 dogbert17 left