<soundnfury>
I'm one of those crazy peroxide advocates... and even *I* wouldn't touch perperoxide with a ten-foot sterilised, passivated, stabilised, anodised pole.
<lamont>
send a grad student to touch it
<soundnfury>
I'm in industry, not academia, so I don't have grad students...
<soundnfury>
... I just have minions ;)
<lamont>
the intern?
<soundnfury>
okay technically just the _one_ minion, and he's not _officially_ my minion
<awang>
If I have a static constructor in an inner class
<awang>
Is that guaranteed to run before the static constructor of the outer class?
<soundnfury>
(and no, not an intern)
<awang>
...Am I even allowed to have a static constructor in an inner class?
<taniwha>
awang: easy way to find out :)
<taniwha>
try it and see :)
Hypergolic_Skunk has quit [Quit: Connection closed for inactivity]
<awang>
taniwha: Well, I tried a static {} block
<awang>
Got a compiler error
<awang>
But I wasn't sure if there was something different I needed to do
<taniwha>
that's not how to do a static constructor, I think
* taniwha
checks
<awang>
Well, static something or another
<awang>
Whatever runs once
<awang>
I'm using a static factory at the moment, but that's racy
<awang>
IT's a known issue, just no one has merged the PRs yet
<awang>
Bornholio's probably because there's a bunch of stuff in there, and there are conflicts
<awang>
idk about this other guy
<Starwaster>
so apparently either SSTU players don't use Deadly Reentry or they don't use SSTU's heat shields... or they would be dying and complaining about it
<SpecimenSpiff>
There have been plenty of complaints about heat shields on reddit
<SpecimenSpiff>
but people have been blaming it on DRE or bad installs
<SpecimenSpiff>
I was having problems on my last install, solved by moving the heat sliders way down
<Bornholio>
yeah i need to scrap my pr and redo it on the new dev no-one pushed it so now its way off on commits
<Bornholio>
saw fixes to rd-108 that didn't fixit
<Bornholio>
and those got pushed
ferram4_ has quit [Read error: -0x1: UNKNOWN ERROR CODE (0001)]
SpecimenSpiff has quit [Quit: Web client closed]
ferram4_ has joined #RO
<Pap>
o/ taniwha
<Pap>
I hope it is treating someone well! I don't even have a current install :(
qwertyy__ has joined #RO
<taniwha>
Pap: :/
qwertyy_ has quit [Ping timeout: 198 seconds]
<Bornholio>
pap thank you
<Bornholio>
question was, should biosample be reusable
<Pap>
Hmmmm, I would say no
<Pap>
But was it before?
<taniwha>
so would I
<taniwha>
#1 rule of the science lab: never return to stock
<taniwha>
(well, ok, #1 might be wear safety gear, but still... :)
<Bornholio>
nuffin in sheet says if an experiment is rerunable after transmit
<Bornholio>
AJE Patches in master today
<Starwaster>
SpecimenSpiff link please?
<Starwaster>
also thanks for telling me when you had problems so I could look into it. appreciated.
<Bornholio>
starwaster is that lossexp -40000 important :)
<Starwaster>
yeah
<Starwaster>
it's supposed to be negative, yes
<Starwaster>
if it's not then bad things happen
<Bornholio>
is that why my pods blew up last time
<Starwaster>
were they positive?
<Bornholio>
was using release 7.2.0
<lamont>
if a base class defines like the ‘+’ operator between two base class objects
<Bornholio>
sorry 7.6.2
<lamont>
if you inherit from that base class, does child + base work because child is-a base (you havent overridden anything)?
<lamont>
and what happens if you have an expliict cast from base to child?
<Starwaster>
bornholio thing is that some loss exp WERE positive in one of the releases and I fixed it. BUT.... it's part of a fallback config that only kicks in if you don't have RO AND you have RSS
<Bornholio>
k
<Starwaster>
if you have RealismOverhaul, even just an empty FOLDER that says RealismOverhaul then the fallback config won't affect you
<Starwaster>
BUT
<Starwaster>
BUT!!!!!
<Bornholio>
but?
<Bornholio>
.meekvoice
<Starwaster>
the main suspect in people's shields not protecting them is Module Manager. Specifically having TWO or more versions of Module Manager. It results in having more than one Physics Globals
<Starwaster>
one of them the stock PHYSICSGLOBALS which will f*** your shit up for you
<Starwaster>
if you're playing RO/RSS
<Bornholio>
yeah need a batch that runs and crushes extra copies
<Starwaster>
so if you know anyone whose shields keep exploding when playing with RO then check for multiple MM copies
<Starwaster>
DRE might not even be necessary in that mix but it obviously makes it easier for your parts to explodey wodey
<Bornholio>
hey do you have a dll for current that has temp range display enable like the test one before
<awang>
I like how my OS says KSP is eating up 20GB
<awang>
But memgraph says ~4.7
<Vader111>
24gb for med at launch
<Bornholio>
awang how much video memory do you have
<awang>
Bornholio: 1GB, I think?
<awang>
NVidia GeForce 650M IIRC
<Bornholio>
memgraph only shows unity specific space, and not preload textures and others
<Bornholio>
while used memory includes all running programs and video shadow copy
<awang>
Ah
<awang>
That makes sense
<awang>
Dangit
<awang>
Spend an getting KSP to start
<awang>
Find out KRASH didn't load because I used "{}" in a format string instead of "{0}"
<Pap>
NULL is bad awang
<awang>
Yes
<awang>
Yes, it is
<awang>
The world needs more strong typing
<awang>
And compile-time stuff
<awang>
{} instead of {0} should be a compile-time check
<awang>
Or at least a warning
TM1978m has joined #RO
ferram4_ has quit [Read error: Connection reset by peer]
ferram4_ has joined #RO
blowfish has joined #RO
Mike` has quit [Ping timeout: 182 seconds]
Mike` has joined #RO
SpecimenSpiff has joined #RO
<SpecimenSpiff>
Has mechjeb been fixed to work with Principia yet?
Moistmelon has quit [Ping timeout: 186 seconds]
SpecimenSpiff has quit [Quit: Web client closed]
awang has quit [Killed (NickServ (GHOST command used by awang_))]
awang_ has joined #RO
ferram4_ has quit [Read error: Connection reset by peer]
ferram4_ has joined #RO
qwertyy__ has quit [Ping timeout: 186 seconds]
blowfish has quit [Quit: Leaving]
NathanKell|AFK is now known as NathanKell
<NathanKell>
o/
<Qboid>
NathanKell: awang left a message for you in #RO [01.02.2018 13:26:33]: "Wow, that's a lot fewer than I expected. Yeah, guess I'll take a look after KRASH, since RealScience seems to be on hold while Pap gets regular science sorted out"
<taniwha>
NathanKell: o/
<NathanKell>
!tell awang* that would be great!
<Qboid>
NathanKell: I'll redirect this as soon as they are around.
<lamont>
^ PEG13 has principia fixes but its complicated: show on navball + impusive burn + internal frame + show patchyconic lies and then manually warp to just before the node for precision. then make sure the patchy conic prediction is aligned with the n-body prediction near the burn when you are in the internal reference frame. then execute it.
<lamont>
hopefully soon it will be patched for reals so you just configure it in the flight planner and press ‘go’
ProjectThoth has quit [Quit: +++out of cheese error+++]
* BadRocketsCo
waves at soundnfury
<Vader111>
lamont did you fix the null ref spam in the vab
<Vader111>
150mb log files are not fun
<lamont>
tried to
<Vader111>
ok i will test it later
<lamont>
there’s a “Force Disable PEG” button now as well
<NathanKell>
You'll have to fill me in on the "tapdance" tomorrow :D
<NathanKell>
I gotta log off
<NathanKell>
o/ all!
NathanKell is now known as NathanKell|AFK
<lamont>
NK: there’s no tapdance if you’re not using Principia
<lamont>
i would not suggest using Principia yet
Vader111 has quit [Quit: Leaving]
Vader111 has joined #RO
Moistmelon has joined #RO
CommandoDiamond has quit [Quit: Leaving]
<TM1978m>
how is every one this morning?
BadRocketsCo has quit [Quit: Bye]
<probus_>
Doin' good. How about you TM1978m?
<TM1978m>
just hopeing we get to release with the Ro/rss/Rp soone... I teend to suck at keeping track of all the depedinces manually
probus_ has quit [Read error: Connection reset by peer]
<TM1978m>
and watching nats Rusty RP-1 vidoes
probus_ has joined #RO
<TM1978m>
just hopeing we get to release with the Ro/rss/Rp soone... I teend to suck at keeping track of all the depedinces manually
<TM1978m>
and watching nats Rusty RP-1 vidoes
Vader112 has joined #RO
Vader111 has quit [Ping timeout: 383 seconds]
BadRocketsCo has joined #RO
<BadRocketsCo>
Howdy
probus_ is now known as Probus
<Probus>
o/
<BadRocketsCo>
How're things?
<Probus>
Goin good. LinuxGuruGamer found me a new version of the Yonge Tech Tree Editor that is a lot easier to use.
<Probus>
How about you?
<BadRocketsCo>
Pretty good :)
<BadRocketsCo>
Going to launch some venus coms sats soon
<Probus>
Hotness! My Venus missions have not been very successful.
Vader112 has quit [Read error: Connection reset by peer]
<BadRocketsCo>
Yeah, my lander just lost conncection on landing, so I'll get more coms sats up
<BadRocketsCo>
Plus I forgot to put the parachute with kevlar, so it would've burned up anyway
<TM1978m>
I hoping this drops on cKan soo... I really suck at manually keeping track of all the depeidnicys lol
<TM1978m>
but having fun watching nat's rusty RP-1 vedeos :P
<TM1978m>
was it just my eyes or did it like like nat had a V-2 built?
BadRocketsCo has quit [Quit: Bye]
Majiir has quit [Quit: CUT THE HARDLINES!!]
VanDisaster has quit [Read error: Connection reset by peer]
Snoozee has joined #RO
Snoozee is now known as Majiir
SirKeplan has quit [Quit: Bye!]
SirKeplan has joined #RO
probus_ has joined #RO
Probus has quit [Ping timeout: 383 seconds]
BadRocketsCo has joined #RO
probus_ has quit [Quit: Leaving]
qwertyy_ has joined #RO
qwertyy has quit [Ping timeout: 207 seconds]
Probus has joined #RO
awang_ has quit [Ping timeout: 182 seconds]
BadRocketsCo has quit [Ping timeout: 182 seconds]
awang has joined #RO
BadRocketsCo has joined #RO
<Maxsimal>
o/
<Qboid>
Maxsimal: NathanKell left a message for you in #RO [01.02.2018 04:03:47]: "yep it was an altitude breaker that did it somehow I think"
<Qboid>
Maxsimal: NathanKell|Twitch left a message for you in #RO [01.02.2018 04:09:24]: "the other deeper problem with the SR contracts is that by the end they pay better than orbital launches. They really shouldn't be paying 1/3 of a lunar launch (and I'm probably not even at the end of the progression string)"
<Qboid>
Maxsimal: NathanKell|Twitch left a message for you in #RO [01.02.2018 04:16:20]: "correction I think I am at the end of the progression. But still, ~50k advance and 50k payout is quite a lot"
<Maxsimal>
!tell NathanKell: Yeah, possibly the top end is too much - although I would stay that the class of LV you need for a light unmanned lunar orbiter and the top end of SR is *very similar* (since both basically require ICBM-class hardward), and I was scaling my rewards based off of my estimated LV costs, not mission complexity, which maybe should play a bigger role
<Qboid>
Maxsimal: I'll redirect this as soon as they are around.
<Maxsimal>
!tell NathanKell: I do have a few ideas now on how to improve the SR contract variety and pacing though
<Qboid>
Maxsimal: I'll redirect this as soon as they are around.
<Maxsimal>
!tell NathanKell For one thing, I think the current thing where all the contracts are 3 types, but its invisibly scaling them up as you do more difficult missions doesn't help a player sense of progression, + the tuning has to fit a curve over the whole range, so that's one issue, I was thinking of splitting the contracts up into a dozen or so SR classes that fit historic mission profiles (with the top end being more like MRBM/IRBM/ICBM tests
<Qboid>
Maxsimal: I'll redirect this as soon as they are around.
<Maxsimal>
!tell NathanKell And then it might be fun to have some contract variants - I came up with 3 I know can be done with CC - whether or not the payload had to be recovered, whether or not the launch window is very time sensitive (you get one shot at it) and whether or not the payload is very g-sensitive so you need a gentle boost phase. I have some other ideas but I'm not sure if they're possible with CC.
<Qboid>
Maxsimal: I'll redirect this as soon as they are around.
BadRocketsCo has quit [Ping timeout: 182 seconds]
<awang>
Does anyone mind checking whether some mod code compiles?
<Qboid>
awang: NathanKell left a message for you in #RO [02.02.2018 06:11:17]: "that would be great!"
<awang>
Just toss this into your regular dev environment
<awang>
It compiles in VS for mac, but doesn't on this online REPL I found
BadRocketsCo has joined #RO
<awang>
So I'd like to know why, at least
<awang>
idk if it's because I'm still thinking in C++ template mode
<awang>
And I don't want the Unity runtime upgrade to break this code
<awang>
The online repl complains "Type `T' does not contain a definition for `GetModuleCost' and no extension method `GetModuleCost' of type `T' could be found"
<awang>
Which sort of makes sense?
<awang>
I was hoping that the compiler would hold off on the check until I actually created a RP0Helper with a specific type
<awang>
C++ template style
<awang>
But I guess C# doesn't work that way?
<awang>
Or at least it used to, but doesn't any more?
<awang>
What would the right way of specifying that T will have that method be?
aradapilot has quit [Remote host closed the connection]
aradapilot has joined #RO
<awang>
Actually, never mind
<awang>
I found the problem
<awang>
Forgot GetModulePart was part of IPartCostModifier
TM1978m has quit [Remote host closed the connection]
<Probus>
I hate when that happens. :)
TM1978m has joined #RO
Senshi has joined #RO
BadRocketsCo has quit [Ping timeout: 186 seconds]
Vader111 has joined #RO
<Vader111>
well lamont i have been trying the latest version of mechjeb peg
<Vader111>
the nullref spam is still there
<Vader111>
also is anyone having issues with procedualfairings
<Vader111>
where it will not shape or be able to have a custom shape appiled to it
Garlik has joined #RO
<Garlik>
hey guys
<Garlik>
RP-1, how to know what pod goes with what course name?
<Garlik>
I want to train my crew for MK2 pod, but I dont know the course that corresponds to
<Garlik>
should be a clear way to find this ingame IMO
Vader111 has quit [Quit: Leaving]
stratochief has quit [Remote host closed the connection]
Maxsimal is now known as Maxsimal|Work
lamont has quit [Quit: lamont]
lamont has joined #RO
<lamont>
yeah right now the only real solution to NREs in the VAB is that if you see PEG is enabled in the flight mode that you disable it before you leave flight mode. i deeply do not understand what conditions are arising where it could be getting left enabled.
TM1978m has quit [Remote host closed the connection]
Senshi has quit [Read error: Connection reset by peer]
<awang>
lamont: If you want to just get rid of the NREs without regard to cleanliness you can just go to the throwing function and immediately return if you aren't in flight mode :P
<lamont>
yeah i’ve put that check into fixed update now
<lamont>
i don’t know what is throwing though nobody has actually showed me an NRE...
<awang>
lamont: Pretty sure I posted one a while back
<awang>
lamont: Stack trace was MuMech.MechJebCore.OnGUI, followed by MuMech.MechJebModulePEGController.DrawCSE
<lamont>
ohhhhhhhhhh
<awang>
Vader111 also posted it a bit back
<lamont>
i’ve been looking at totally the wrong thing
<awang>
What were you looking at?
<lamont>
okay well there is an enabled chedck in there, i’ve been trying to make sure it gets disabled properly
Garlik has quit [Ping timeout: 180 seconds]
CommandoDiamond has joined #RO
aradapilot has quit [Ping timeout: 182 seconds]
<soundnfury>
so at work this week I spent two days chasing down a kernel panic that turned out to be because I'd used min() instead of max().
probus_ has joined #RO
<awang>
My condolences
<awang>
What did the min() instead of max() do to cause the panic?
Probus has quit [Ping timeout: 198 seconds]
<soundnfury>
it meant that the bpf runtime thought it only needed to allocate a 32-byte stack, when the program actually used 512 bytes
<soundnfury>
and since the helper function it called failed -EFAULT because reasons, it ended up memset()ing those 512 bytes to zero. Including the return address.
<soundnfury>
Seeing your computer attempt to *execute* a null pointer is not fun.
<awang>
Ouch
<soundnfury>
ah well... nowadays when my code has bugs I just say to myself:
<soundnfury>
"well hey, it's not as bad as Spectre" xD
<awang>
Wonder if the guy(s) who invented branch prediction are still around
<awang>
And what they would say about this entire situation
<awang>
"...You can do that?"
<awang>
"I KNEW it!"
<awang>
idk
aradapilot has joined #RO
<lamont>
doubt it
<lamont>
they were trying to make CPUs faster
<awang>
Are there any CPUs that have branch prediction but aren't vulnerable to Spectre?
aradapilot has quit [Remote host closed the connection]
<soundnfury>
well, from my reading of it, the bugs are serious brown-paper-bag-level bad design
<awang>
Well, besides some of AMDs more recent ones
aradapilot has joined #RO
<soundnfury>
take spectre v2 for instance, the Branch Target Buffer (used to predict indirect branches) should be per-address-space
<awang>
IIRC Spectre was mostly because the entire branch target address wasn't stored in history, just bits in it
<awang>
s/bits in/bits of
<Qboid>
awang meant to say: IIRC Spectre was mostly because the entire branch target address wasn't stored in history, just bits of it
<awang>
BTB per address space sounds awful for cache and/or die space
<soundnfury>
not just for security reasons, but for performance as well (why pollute your BTB with addresses from another context that will almost *never* be correct?)
<soundnfury>
awang: you don't have to make it bigger, you just have to tag entries. It's like PCID in the TLB
<awang>
That's true
<awang>
(for both things)
<soundnfury>
also, it's the branch _source_ address that they only stored part of
<soundnfury>
but even if they'd stored all of it, *with* KPTI there'd still be an attack
<awang>
Oh, it was the source?
<awang>
....How does that make sense for branch prediction?
<soundnfury>
yeah, the source is part of the key it uses to look in the BTB for a target to predict
<awang>
Oh, as a key
<awang>
I thought CPUs would just use the source address as a key into a table, but not store the source address
<soundnfury>
... that's what happened, they used _part_ of the source as a key
<lamont>
that is meltdown though which is specific to intel bugginess… spectre is the one more inherent to branch prediction?
<awang>
So I thought storing the entire branch target would defeat Spectre, since you'd have to actually jump to the target address to get it into the history buffer
<awang>
I thought meltdown was not checking permission bits
<soundnfury>
le sigh... read the paper
<awang>
Permission bits in the TLB, I think?
<lamont>
(got the paper open now trying to decipher… again...)
<awang>
Same here
<soundnfury>
yup, me too ;)
<soundnfury>
meltdown is having a single cache common to the supervisor-mode and user-mode versions of the same address space
<lamont>
the BTB issues seem to be one variant of spectre attacks
<soundnfury>
and then having speculatively-executed accesses populate part of that cache, which can be detected through timing
<soundnfury>
and the *which* part was populated leaks info.
<soundnfury>
BTB is Spectre variant 2
<soundnfury>
which would totally be fixed by tagging BTB entries with PCID and supervisor bit, which like I say you should be doing _anyway_
<soundnfury>
Spectre variant 1 would be somewhat-similarly solved by having cache updates be part of the pipelined state (like register forwards) and only applying them when the insn retires/commits
BadInternetCo has joined #RO
<awang>
Transactional memory for caches?
<soundnfury>
sort of? You have to have that already for the register file
<soundnfury>
and for the *writes* to those caches
<soundnfury>
and all the data hazard stuff (RAW, etc.) is already solved for those purposes
<soundnfury>
basically, if any chips designed after these attacks were published are vulnerable, or require a noticeable slowdown to avoid being vulnerable, then the designers are monkeys and should be fired
<soundnfury>
(the existence of variant 2 in the first place suggests that the designers are, at best, orang-utans.)
stratochief has joined #RO
<lamont>
well the designers may very well have been concerned about these issues but some middle manager above them likely told them it was good enough and to ship it
BadRocketsCo has quit [Ping timeout: 383 seconds]
<soundnfury>
fair point
<soundnfury>
but *someone* in this story is an orang-utan :P
<soundnfury>
also, if you're Roger Boisjoly and a manager tells you to ship it, put your foot down and say no.
ProjectThoth has quit [Ping timeout: 182 seconds]
<lamont>
then you get passed over for promitions and get bad performance reviews until you quit
<soundnfury>
if so, then that's not a company you want to be working at anyway
CommandoDiamond has quit [Quit: Leaving]
<soundnfury>
I maintain that that's part of an engineer's professional responsibility
<soundnfury>
especially if we want people to be able to trust computers (and space shuttles) the way they trust, say, bridges
<lamont>
i agree, but i have no kids and no health problems and i can pick up and switch jobs much easier than some people
<soundnfury>
yes, and when that happens, the lawsuits are meaningful
<soundnfury>
and the engineer who signed off on the design is in deep shit
<soundnfury>
whereas with something like Spectre, no individual is identifiably responsible
<soundnfury>
in your scenario, the manager overrules the engineers so they aren't responsible, and the managers set up the CYA and 'consensus' decisionmaking so that _they_ won't be held responsible either
<lamont>
the engineering who signed off on that bridge did that back in the 1960s, and 40 years of subsequent inspections never turned up the flaw which spreads the fault very thinly
<soundnfury>
(and in 30 years not much has changed, only now fewer people read comp.risks)
<lamont>
sure and that’s the problem, in a company with limited liability you don’t see criminal liability for negligence until the actions become really idiotic
<lamont>
and the engineer who raised the alarm about challenger blamed himself for not being able to convince his management chain, who just didn’t want to hear it, up until nearly the day he died.
<soundnfury>
Right, which is presumably why the engineer isn't in jail for negligent homicide (which I believe is a thing that can happen)
<soundnfury>
^ that was re the bridge btw
<soundnfury>
the challenger engineer, depends which one you're talking about (Boisjoly or Ebeling)
<soundnfury>
Ebeling blamed himself
<soundnfury>
Boisjoly blamed everyone _but_ himself and went on a speaking tour about how he was the noble whistleblower
<lamont>
whoah 1988 comp.risks
<lamont>
uunet e-mail addresses
<soundnfury>
(who, uhhh... stayed quiet during the critical conference call)
<soundnfury>
tasty tasty bang paths :)
<lamont>
thats an e-mail addressing scheme i’ve not seen in a long, long time…
<lamont>
anyone have a link to Mr. Holio’s spreadsheet?