Charcoal HQ

@Andy basically, making sure we test things everywhere >:D

Andy

Tests are good. I like tests.

don't get testy now

@Undo @ArtOfCode codecov is bitching, can i ignore it?

Andy

8:04 PM

@ThomasWard probably

!!pull

yeah it looks like Smokey doesn't care about codecov tests only the travis and circle ones

!!/pull

!!/pull

[ SmokeDetector ] SmokeDetector started at rev ac05a2b (Thomas Ward: Fix obfuscated username issue with displaying 'alias'ed author names.) (running on teward / Midnight Eclipse)

Restart: API quota is 13478.

@Undo that should work now. I forgot about the obfuscated names part when we did the refactor and started using dulwich and the commit information dict my small function generates. Oops.

kinda funny though to see your name as 'U' in the notices though xD

hey lookit a oneliner xD

@ArtOfCode @Undo god I love in-line if statements xD

8:23 PM

CI on 8e0add7 failed.

wat

oh

TBH I like how Circle lists the different test steps it does

in its own sections. Travis, it's not easy to always see immediately what broke :P

> The command "flake8 ./" exited with 1.

CI on 7d034fa succeeded.

In red and bold

@JF Yes but i always assume that after that is the test. Because I'd expect it to complain first then show where it failed.

sorta like circle does with green/red indicators and collapseable and expandable sections

!!/pull

8:27 PM

[ SmokeDetector ] SmokeDetector started at rev 7d034fa (('Thomas Ward', 'teward@users.noreply.github.com'): Flake8 fixes) (running on teward / Midnight Eclipse)

ooops?

Restart: API quota is 13265.

\o @('Thomas Ward', 'teward@users.noreply.github.com')!

can i taser you :P

Please don’t :)

8:32 PM

wow, it returns a tuple, not a list. I thought it returned a list. Oops. (Updated)

CI on 56a1dd8 succeeded.

don't pull yet

@QPaysTaxes licks ice cream cone

CI on 55eaf90 succeeded.

!!/pull

[ SmokeDetector ] SmokeDetector started at rev 55eaf90 (teward: Add myself to censored committer names.) (running on teward / Midnight Eclipse)

8:36 PM

yay it worked.

@ThomasWard Why do you censor those names?

Restart: API quota is 13188.

@JF Undo's has always been censored. I censored my own name to make sure my oneliner worked.

Ah, OK.

and since it did i'm removing my own name.

8:38 PM

CI on 685b05b succeeded.

!!/pull

[ SmokeDetector ] SmokeDetector started at rev 685b05b (Thomas Ward: Remove teward from censored committer names.) (running on teward / Midnight Eclipse)

yay!

Restart: API quota is 13155.

@JF as for why his name is censored, well, you'll just have to ask @Undo

maybe so people know who in here actually committed?

Chaos Incarnate.

oh wait, that's one of mine, nevermind

8:42 PM

@QPaysTaxes Whose GitHub name?

@NobodyNada @Undo

it's Undo1

no we mean the name he commits as, not his username

like where it shows my name, Thomas Ward, that's what I have it set to in git

i don't know what his is set to

[ SmokeDetector | MS ] Username similar to website in answer: Regular expression to match DNS hostname or IP Address? by Thangaraj on stackoverflow.com

@ThomasWard His real name

8:44 PM

fp- by J F

@ThomasWard Undo1 from a computer, his real name from GitHub.

[ SmokeDetector | MS ] Repeating characters in title: HIIIIIIIIIIIIIIIIIIII CHRISSSSSSS HOLA AMIGO :) by CHRIS on biology.SE

tpu- by tripleee

Ferrybig

^ looks like homework to me, look at the last sentence

Conflicting feedback on HIIIIIIIIIIIIIIIIIIII CHRISSSSSSS HOLA AMIGO :).

It's that guy again

8:47 PM

fp- by NobodyNada

yes

i flagged for their mods

right

but that's what we meant

have you already flagged it?

the first one was by a user BOB so technically this is the second time CHRIS posted it

CI on 5911606 succeeded.

!!/pull

...the Raspberry Pi 3 has a 64-bit processor, right?

8:48 PM

[ SmokeDetector ] SmokeDetector started at rev 5911606 (Thomas Ward: Fix 'escape character invalid' warning on Travis) (running on teward / Midnight Eclipse)

Restart: API quota is 13072.

@NobodyNada yes but armhf vs. arm64 is negligible difference so don't bother going after arm64

So why is new returning an 8-byte aligned pointer?

@NobodyNada uname -a

> The address of a block returned by malloc or realloc in GNU systems is always a multiple of eight (or sixteen on 64-bit systems).

BOB 1 BOB2 CHRIS 1 CHRIS 2 just now

8:49 PM

gnu.org/software/libc/manual/html_node/…

@NobodyNada what does uname -a say about your system

SHOW

GIVE DATA

Linux qcpi 4.4.50-v7+ #970 SMP Mon Feb 20 19:18:29 GMT 2017 armv7l GNU/Linux

@NobodyNada what distro on your RPi?

Raspbian

also, what does arch say?

8:51 PM

armv7l

you're probably not on arm64, and using the older 32bit builds - probably armhf because Raspbian arm64 would only work for the Pi 3 and the Raspbian image is identical for Pi2 and Pi3

that's why you're seeing the older 8bit addressing

Ohhhhhh

But the compiler is assuming 64-bit

This line is returning an 8-bit aligned pointer, but the compiler is assuming it's 16-bit aligned

@QPaysTaxes Probably, but I don't know

@NobodyNada your system can definitely report a different arch on the processor itself than what it would be normally

but i don't know how your compiler is set up

and I'm not a C/C++ guy

So...I'm using 32-bit libraries but compiling 64-bit code?

vice versa probably

I don't know enough about Raspbian to really answer

Stackoverflow or somewhere might be more useful

8:55 PM

Maybe

So what's going on is that that new I linked is returning an 8-byte aligned pointer

but that breaks this code:

Ferrybig

I use a old raspberry pi as server, I can help testing

That is NOT what I meant to copy and paste

@QPaysTaxes In a bit

B+ │0x76f1a398 <allocateSideTable+44> bl 0x76ceed98 <_Znwj@plt> │
>│0x76f1a39c <allocateSideTable+48> mov r10, r0 │
│0x76f1a3a0 <allocateSideTable+52> add r8, sp, #8 │
│0x76f1a3a4 <allocateSideTable+56> vmov.i32 q8, #0 ; 0x00000000 │
│0x76f1a3a8 <allocateSideTable+60> mov r1, r10 │
│0x76f1a3ac <allocateSideTable+64> sub r0, r5, #4 │

Let me prettify that

sorry i can't read assembly :p

@ThomasWard The problem is the last instruction: vst1.64 {d16-d17}, [r1 :128]. I think the [r1 :128] indicates r1 is 128-bit aligned (i.e. 16-byte aligned)

But the malloc is returning a 8-byte aligned pointer, as it would on a 32-bit system

That's why I think I'm using a 32-bit libc with 64-bit code

Heh. Housemates group chat: "so when's everyone getting back from holiday?" <insert various responses of either saturday or sunday>

me: "so who got back today then?"

housemate: "me, but how the !&*$ did you know?"

this being the advantage of OpenDNS telling me exactly when anything queries to anything other than root-servers.net

9:06 PM

I hope it doesn't subject you to the contents of the actual queries :P

I hope you thoroughly explained this to your housemates ... they'll think you are the NSA

@Undo it'll tell me which domains devices on the network have sent DNS queries for

@tripleee plot twist: I am the NSA

@ArtOfCode has to be... interesting

I could get the request contents if I wanted to flash custom firmware onto the routers, but I'm not sure I can be bothered, honestly

@Undo amusingly, I can also block categories of websites with custom messages. One guess how I used that one.

Experts Exchange, I hope

Yes

9:09 PM

@Undo yeah, I'm not that high-brow

You might consider what kind of legal liability you get into doing that. Having an active way to see what folks are looking at might void some of the normal ISP-type protections.

Yep, testing shows malloc returns 8-byte aligned pointers, not 16-byte aligned

:\

I can't believe that was causing a problem

also, someone tell me what's special about numbers between 2774 and 2776? I'm getting that many requests a day to root-servers.net

I.e. "Some guy looked at illegal stuff on your network" turns into "Some guy looked at illegal stuff on your network and you didn't report it"

works out to a request every 31 seconds

9:11 PM

@ArtOfCode is that the actual distribution pattern?

@Undo offloaded onto OpenDNS themselves, because I blocked "illegal stuff" categories

@tripleee not a clue, it's not that fine-grained

@QPaysTaxes All to root-servers?

@ArtOfCode Across how many devices?

I'd bet it's some 30s increment

I was going to say 29

@Undo there should be zero on the network for that period, excluding three routers

@QPaysTaxes The problem, however, is that clang is assuming it returns 16-byte aligned pointers.

It's assuming that because it's supposed to be guaranteed to return 16-byte pointers, in the GNU standards

Thomas said it's because I'm using a 32-bit libc on a 64-bit system

Because Raspbian is 32-bit

9:14 PM

well, and 50m of Cat 6, but I doubt that's generating requests somehow

one of those new-fangled DNS caching cables

@Undo of course, how could I miss that

Okay, this is even stranger (and apparently it is that fine-grained, @tripleee):

April 18th by hour

9:33 PM

@QPaysTaxes those are DNS requests

all to a.root-servers.net

and all A requests, at that

So it looks like Clang is not generating 64-bit code, but for some reason, it is assuming 16-byte memory alignment from malloc.

And O.o someone actually used DSS

9:49 PM

FWIW

i'm gonna have to shut off Midnight Eclipse briefly

I need to restart my hypervisor with the power rerouting

!!/location

@ThomasWard teward / Midnight Eclipse

!!/stappit

[ SmokeDetector ] SmokeDetector started at rev 5b07171 (SmokeDetector: Auto blacklist of tl900\Wheadlamp by tripleee --autopull) (running on ArtOfCode/EC2)

!!/alive

@JF plz send teh coffee

9:50 PM

!!/location

@JF ArtOfCode/EC2

!!/pull

[ SmokeDetector ] SmokeDetector started at rev 5911606 (Thomas Ward: Fix 'escape character invalid' warning on Travis) (running on ArtOfCode/EC2)

Restart: API quota is 16226.

10:11 PM

[ SmokeDetector | MS ] URL in title: how to investigate diff between theguardiam.com and tehguardian.com by user567324 on security.SE

fp- by J F

lol for slow boots xD

@ArtOfCode Midnight Eclipse should be coming back up shortly.

10:29 PM

@SmokeDetector Thanks Smokey, that prompted me to write my first answer on security.stackexchange.com :)

what's the purpose of the body_summary field in a rule?

@ArtOfCode Eclipse is comoing back online

[ SmokeDetector ] SmokeDetector started at rev 5911606 (Thomas Ward: Fix 'escape character invalid' warning on Travis) (running on teward / Midnight Eclipse)

!!/standby ArtOfCode

Restart: API quota is 12563.

[ SmokeDetector ] SmokeDetector started in standby mode at rev 5911606 (Thomas Ward: Fix 'escape character invalid' warning on Travis) (running on ArtOfCode/EC2)

10:38 PM

new rule proposal #662

I'm thinking some of the really big regexes could be refactored in a similar fashion

or look at the Indian cities list, that's a useful tool if you want static strings, but it wouldn't work here because the fragments are regexes, not literal strings

(spent a fair amount of time trying before I figured that out ...)

The city list could be something to store in the DB.

there's a large number of lists of phrases and other fragments, dunno if the database is the right place but they could definitely benefit from some gentle refactoring

11:07 PM

@QPaysTaxes DEBIAN_FRONTEND=noninteractive?

@QPaysTaxes That's a trick question. All of them are unique in some way :)

All right - trick statement

ArtOfCode/EC2 received failover signal.

Restart: API quota is 19914.

[ SmokeDetector | MS ] Bad keyword with email in answer: RecyclerView showing nothing by Jean Spencer on stackoverflow.com

that's my fault I forgot to install some hardware in that hypervisor so blah

gotta install the hardware and reboot

11:25 PM

[ SmokeDetector | MS ] Mostly dots in answer: Proof that the following multiplicative groups modulo m are cyclic by s n on math.SE

tpu- by DavidPostill on RecyclerView showing nothing [MS]

@QPaysTaxes Use travis encrypt to secure a personal access token from GH settings, then git clone https://$GH_TOKEN@github.com/user/repo, then commit on that and push.

naa- by DavidPostill

@QPaysTaxes Affirmative.

@QPaysTaxes Yes, if you count me as "public"

[ SmokeDetector | MS ] Bad keyword with email in answer, blacklisted user: Convert python to Java opencv code by Jean Spencer on stackoverflow.com

tpu- by 4castle