« first day (1218 days earlier)      last day (2595 days later) » 
00:00 - 05:0005:00 - 00:00

12:03 AM
by the way flake is complaining
 
@quartata I just sent a post down the socket, if you got it
@quartata when is it not
 
> error from callback <function on_msg at 0x7f1890a19bf8>: 'type'
Does it not have a type?
 
no
 
key that is
darn
 
not from PostsChannel
 
12:06 AM
try again
cool thx
ugh that's gonna be a nightmare to parse
 
It's HTML. What are you trying to do? There's probably an easier way to do it, like sending post data down the API socket.
Anything that isn't ApiChannel is not intended for consumption outside metasmoke
 
I just want to be notified when a post is reported
this will work though
 
@quartata The API channel does that already: you'll either get a flagged or a not_flagged messagae
 
That's for autoflagging?
Oh.
 
yes, but it contains post details
 
12:09 AM
the WS docs could use a rewrite then
 
probably
 
Can you send a post to API?
 
but then again 90% of our documentation is people anyway
@quartata hang on, need to figure out what that sends
 
90% of our documentation is in chat where you can't find it unless if you were there at the time
 
"post things" is this, @quartata
 
12:14 AM
ahhh much better
how do I get the MS id?
I mean I can do by-post
I guess
 
that was the not_flagged variant; you get something slightly different from the flag_log message
@quartata should be under post
yeah, not_flagged.post.id
 
{'message': {'not_flagged': {'post_link': '//stackoverflow.com/a/34978634340834', 'post': {'post things': True}}}
@ArtOfCode oh on a real message?
 
2 mins ago, by ArtOfCode
"post things" is this, @quartata
 
ok
sorry for the confusion
 
I just couldn't be bothered to go get all the post details to accurately reconstruct a real message :P
 
12:28 AM
[ SmokeDetector | MS ] Manually reported question: I'm a perm IT recruiter in downtown Boston looking for a developer by bebchi on stackoverflow.com
 
@SmokeDetector k
 
Yo
 
@JasonC o/
 
@SmokeDetector Idea for rule: dots make up > 40% of text
 
12:39 AM
@angussidney on it
 
Cool, thanks
 
sd - k
 
!!/blacklist-keyword n\.u\.m\.b\.e\.r
 
@NobodyNada You don't have code privileges, but I've created a pull request for you.
 
12:49 AM
!!/test n.u.m.b.e.r
 
> Would not be caught for title, body, and username.
 
one second one second I gotta make a pull request
 
!!/test n.u.m.b.e.r
 
> Bad keyword in title
----------
Title - Position 1-10: n.u.m.b.e
 
I probably made Flake unhappy earlier
not enough sacrifices, clearly
sorry flake
 
12:51 AM
yesterday, by ArtOfCode
Turn thee thrice ninety degrees anticlockwise, then flip thee head over heels backwards. Proceed to sacrifice thine firstborn child at the altar of Flake, and it shall forevermore accept thine code.
Lies!
 
Ah, I see where I went wrong. I flipped me heels over head forwards.
fixed
 
CI on a403f4a failed.
 
-_-
 
@ArtOfCode nice joke :P
 
now you're just being vindictive flake
 
12:54 AM
Thee shall never overcome our overlord, Flake
 
...you're kidding me. W291.
 
CI on 41a882b succeeded.
CI on 41a882b succeeded.
 
@angussidney did you need that PR for something, by the way?
@SmokeDetector heh, CI race condition. Nice.
!!/pull
 
[ SmokeDetector ] SmokeDetector started at rev 41a882b (ArtOfCode: TRAILING WHITESPACE MY FOOT) (running on ArtOfCode/EC2)
 
@ArtOfCode testing, but you can merge it and I'll create a new one if I need it
 
12:56 AM
Restart: API quota is 18753.
 
I'm making a US which can toggle autopull on or off for a commit
 
Ah
 
CI on 1581e6d succeeded.
 
Have any of you guys ever written an SE API based desktop application that used the implicit OAuth flow?
 
@JasonC I've done those two things separately, but not together
 
1:03 AM
Yeah, it's the "together" that I'm losing sleep over. I'm trying to figure out if there's a way I can do it without a crappy looking can-you-really-trust-me embedded browser. I'm curious if anybody has ever come up with another smooth way to integrate it.
It's too bad there's no widely-accepted API to communicate with web browsers.
 
@JasonC I think there is, but I don't know off the top of my head. Lemme look at some docs.
@JasonC Huh, apparently SE suggest using an integrated browser. The only other way I see of doing it is to additionally host a web server, and use the redirects to store tokens there.
 
I mean, I suppose I could give a user an option of doing it in their browser then copy + pasting an access token, in addition to a fully automated embedded browser.
 
actually, assuming you have a server, hosting your own web server would be a decent way to go about it
 
Yeah I've been looking through the API docs, it's annoying.
I was thinking about that, but, something doesn't feel quite right about adding another server for the desktop application to depend on, when it only really needs to be able to get in touch with SE. If my server goes down, the application stops working.
Thanks for looking btw.
 
@JasonC Yeah, the dependency is annoying, but the issue of maintenance can be dealt with pretty easily. Make it a dead-lightweight Node server that only responds to the redirect_uri. In the app, you generate a UUID and send that to SE as the state param; that gets passed to your web server; you store the access token against the UUID, and then your app can fetch it using the UUID.
What are you thinking of making, btw?
 
1:16 AM
An app to download and view, offline, all of the content you've posted to all of the sites.
 
@JasonC Does that need authentication?
 
Not necessarily, but then there's the potential for API quota issues.
I'd rather users be responsible for their own quotas.
 
Also true
And it looks like you can't easily search answers by author ID, anyway
 
Yeah it's all way easier with the API.
And I'm not trying to get anybody rate limited with page scraping requests, either.
 
@JasonC Yeah, that's what I meant. You can use /search/advanced in the API to look for questions by author ID, but not for answers.
 
1:21 AM
You can use /posts.
/users/id/posts, or /me/posts
 
@ArtOfCode what's the latest time that you are usually online?
 
@angussidney It's a Saturday, so... pretty late early
 
@ArtOfCode on a normal day?
 
I'd already be gone :P
1 latest, normally. I aim for 12, but it usually doesn't work out
 
Cool
 
1:23 AM
@JasonC bah, so you can, though the latter needs auth
so yeah, auth would be easier
also in today's revelations: writing tests is boring
 
That's why you should never test anything.
 
that's the usual strategy, but for a framework that two scripts already rely on, and potentially more could in the future...
 
0
Q: Origin of "turn thrice widdershins"

bwDracoI've seen "turn thrice widdershins" (or variations thereof, such as "turn around thrice widdershins") several times online (1, 2, 3). It seems to be Gaelic in origin, having to do with fairies, but I can't find any concrete information about where this phrase actually came from. Any idea as to t...

(cc @ArtOfCode)
 
@bwDraco Widdershins: in a direction contrary to the sun's course, considered as unlucky; anticlockwise.
 
I know what "widdershins" means, but I don't get the significance of the phrase as a whole.
 
1:29 AM
shrug
2
 
@angussidney sorry for the delay, had to eat
Hey look at that, Flake likes me.
Speaking of which that post still needs flags which boggles the mind
It's been there for 47 minutes
 
@quartata can't review it properly ATM, will have a look later. Thanks!
 
no problem
cc @ArtOfCode if you want to review
 
sure
 
Yeah I can't wait until we port to Py3
 
1:38 AM
gonna be hard
tests'll explode in our face
 
A nuisance truly
 
but we actually hard-forced py2 now so THERE
 
CI on ad14da7 succeeded.
 
and yes I want Art's review. And probably Undo's on your filter, @quartata but eh
 
He reviewed it already, my message was in reply to one of his comments
@SmokeDetector ding
Another reason I disabled on PPCG
 
1:40 AM
!!/rev
 
Since I'm sure there's an esolang that's all dots
 
LOL @ArtOfCode
@quartata there's several
 
Then again I turned on stripcodeblocks....
ehhh
eh
I'm sure it'll be fine
We can enable it there if it's a problem
 
watches it all explode, then chastises @quartata
 
1:41 AM
@SmokeDetector haha
 
!!/blame
 
@quartata It's Thomas Ward's fault.
 
then cleans up the mess left over from flaming SmokeDetector parts
 
I don't think so @ThomasWard
!!/location
 
@quartata ArtOfCode/EC2
 
1:43 AM
Safe to pull?
 
!!/pull
 
[ SmokeDetector ] SmokeDetector started at rev ad14da7 (ArtOfCode: Merge pull request #602 from quartata/master) (running on ArtOfCode/EC2)
Restart: API quota is 18620.
 
>>> findspam.mostly_dots(".@.1.8.4.4. .7.8.2. .8.0.9.6. .s.k.y.p.e. .s.u.p.p.o.r.t. .n.u.m.b.e.r. .s.k.y.p.e. .t.e.c.h. s.u.p.p.o.r.t. .p.h.o.n.e. .n.u.m.b.e.r.", None, None)
(True, u'Post contains 67 dots out of 134 characters')
good
 
@quartata do me a favor and run some private-range IP addresses through your detections. Just saying.
 
Are you going to have a question of all IP addresses?
 
[ SmokeDetector | MS ] Manually reported answer: How to install Javassist to an Ant project by Andana on stackoverflow.com
 
no but our phone number detection triggered on an IP address that was 10 digits long, so I want to try and see if the FP risk here is just as evil
 
Nah, this is based on the ratio. So it would have to be a tiny question with a lot of IP addresses
4 dots to 30 characters won't trip it.
But this will:
Hopefully people don't have IPs like that.
 
blah i'm tired
someone hit me please
 
@ThomasWard pow
Anyways if there's an FP I can increase the threshold but I'd be pretty surprised if this trips on something that isn't weird anyways
 
1:52 AM
@quartata how does it do with this answer?
contains 331 IP addresses
 
Too long to paste in chat.
 
true, and that's quite the edge case
 
But with only 331 IP addresses it would have to be only 3310 or more characters.
Also code blocks are stripped
so any IP addresses in there wouldn't count
There's an easy way to find out: bump it
 
!!/test best service
 
> Would not be caught for title, body, and username.
 
2:01 AM
!!/test diabazole
 
> Bad keyword in body, bad keyword in title, bad keyword in username
----------
Title - Position 1-10: diabazole
Body - Position 1-10: diabazole
Username - Position 1-10: diabazole
 
yeah it can't delete ancient :P
but we can move them ;)
 
Installing WSL at 2AM, what could possibly go wrong
 
@ArtOfCode evil
 
@ArtOfCode if you were on my internet, you wouldn't be going to sleep tonight :P
 
2:12 AM
@angussidney Then I guess it's a good thing I'm running at 85 Mbps
 
8.5 on a good day... =(
 
@ArtOfCode I have you beat at over twice that amount.
99% of the time.
 
okay so apparently that worked without breaking at all
 
it's always surprising when that happens
 
...apparently your home directory is not the same as your Windows home directory. That's confusing.
nothing a usermod won't solve, though
 
2:30 AM
@ArtOfCode No, but you can get to your Windows directory via /mnt/c/Users/YOURUSERNAME/ and it's 777'd
 
@ThomasWard aye, I figured that, and set that as home instead :P
 
@ArtOfCode could you pull some stats on queue timings? once you get SSH to work with WSL of course :)
 
that reminds me
do we want Queue Timings synced up to the git repos or no?
i have the pickle in gitignore for now but...
 
Nah, leave it in gitignore
Stuff that changes frequently based on what is hosting the bot should be gitignored
e.g. user blacklists, queue timings
 
2:33 AM
yep.
 
Though we should try making a solution to sync notifications etc with MS between hosts
 
^ disregard that CI, I am in OCD mode sorta, and wanted to get the comments in line with each other. (It really irks me heh)
@angussidney have MS store a copy of the pickle data dict?
send that down to Smokey at startup of the socket maybe?
 
CI on 4b53165 succeeded.
 
^ disregard that CI i mean
 
I might give that a try once I eventually finish learning Rails
FR: Give CI status on the userscripts and website repos too
 
2:35 AM
well MS just needs to be able to return the dict and then that needs parsed by Smokey. And that can be stored in the pickle, but the core issue is it has to be a dict for the software to use it heh
 
I'm going to make an issue for that
@SmokeDetector why
 
@angussidney Body - Position 1053-1128: n<a href="https://i.imgur.com/titSsYZ.jpg" rel="nofollow noreferrer"> </a>t
 
@SmokeDetector tpu
 
@ThomasWard Blacklisted user and registered question as true positive.
 
2:36 AM
(it's an NSFW image)
 
tpu- by angussidney
 
These is an invisible link to a NSFW image in this post, please flag as spam — angussidney 17 secs ago
@angussidney s/these/there/
 
[status-fixed]
 
OP rejected the edit
 
figures
 
2:38 AM
I'm 2k, should I edit?
 
bring a nuke on them
@NobodyNada flag as spam let it die
 
Why are people editing spam?
 
someone else edited
 
@angussidney should already be working, but I wonder if @ThomasWard knows why chmod calls do literally nothing in /mnt/c/Users/username/Documents
 
The user posted a comment translating to 'suck my dick'
 
2:39 AM
@ArtOfCode Because /mnt/c/* is an NTFS partition or exFAT and more than likely will not work at all.
@ArtOfCode for such non-UNIX-perms-supporting directories, the umask can only be changed as a mount option
and can't be 'changed' per file
this is why ext4 and Windows don't get along well.
 
@ThomasWard NTFS. FFS, that's no use Microsoft
 
@ArtOfCode basically, the Windows partition formats don't support UNIX-style permissions, so the umask is set in the mount options and can't be overridden because there's no support for the UNIX permissions styles
 
yeah
 
and therefore it just "doesn't work" in any deployment
 
which is totally useless because I want to actually control permissions on those
 
2:41 AM
but you can say the same for FAT/FAT16/FAT32/exFAT as well
 
anyway, queue stats
 
@ArtOfCode this is why the home directory is in its own Linux-compatible location on WSL
 
yeah, that might be a decent idea after all
 
@SmokeDetector aaaand gone
 
finally
 
2:45 AM
Relevant to that invisible image detection by the way: meta.stackexchange.com/questions/292600/…
 
@ArtOfCode thanks
 
Assuming es. and pt. are Spanish and Portuguese SO they look like they have the worst median & count
 
So it turns out that Ubuntu is better at being Ubuntu than Windows is. Who knew?
 
Wow.
 
2:50 AM
LOL
 
Anywho we could add an "all-whitespace link" rule for the time being
Since that's obviously pretty shady
 
@quartata I think the detection on a single character or so is decent enough, but go ahead if you want.
Just make sure to add some unit tests :P
 
@QPaysTaxes haha
 
or better yet ask @ArtOfCode to add unit tests, they love unit tests.
 
I don't have time today but maybe tomorrow.
 
2:51 AM
@QPaysTaxes that's hardly a workaround when I can see the blue link :P
 
@ThomasWard Really?
@ArtOfCode will you do all my unit tests forever please
 
you wish
 
laughs evilly
 
dammit
 
What about "link containing no alphanumeric characters?"
 
2:52 AM
We kind of already have non-latin link for that right
 
I don't think that'd get too many false positives, and it will catch whitespace, punctuation, etc
 
@QPaysTaxes likewise, but I can still tell it's blue and that the link has dQw4w in it
 
@angussidney thanks for updating with the relevant links, I just fast-typed the post and didn't think to include the relevant MS and SmokeDetector project links :)
 
No problem
 
What about "body contains prime number of hyphens and link contains only cat-faced emoji"? It's classic spam.
 
2:55 AM
what about "post is comprised mainly of processed meat-like substance with a gelatinous glaze"? And that's a meta reference, if anyone gets it.
 
> Url in title
----------
Title - Position 6-20: http://foo.com
 
> Url in title
----------
Title - Position 7-21: http://foo.com
 
"Post contains 'mansplaining' and user's avatar is a hamster"
 
I don't think Smokey tests markdown with !!/test
 
2:56 AM
it doesn't.
 
@QPaysTaxes Was that... A/B TESTING? Baaa ha ha ha!
 
does HTML work?
 
:(
 
2:57 AM
Too bad I work from home.
 
@quartata if you wish to create that whitespace only link text rule feel free, I wish you luck.
I'm going to go grab some vodka and put it in my Dr. Pepper. And enjoy by alcoholic soda
 
@ThomasWard we already have one-character link rules and punctuation-only link rules, so no need to create any new ones
 
@angussidney but we don't have two-space link rules
proof of concept is that we can bypass the current issue with more than one space
Unless SE decides to roll out the feature I'm requesting, in which case then our issue is moot.
@ArtOfCode Unrelated...
>>> from __future__ import division
>>> 5 / 4
1.25
>>> 5 // 4
1
 
Holy crap
The division of tomorrow, today
 
in Python 2
 
Py3 division is great
 
@QPaysTaxes ...fine :)
I figured everyone in here would get the reference
and if they didn't, they could just run it in a REPL
 
@JasonC I COME FROM THE FUTURE WHERE DARK CHAOS POWERS ARE WIELDED BY THE MIGHTY! I NOW OWN EXISTENCE AND CONTROL THE FATE OF ALL OF THIS ERA OF TIME! THOU SHALL BOW BEFORE THE FUTURE-KIND THAT IS ME!
maybe I shouldn't have any vodka tonight
 
Don't drink and chat kids
3
 
@ThomasWard ... please step out of the car sir
 
3:06 AM
@quartata *looks out of the viewport of the heavy tank he's driving, then speaks over the megaphone mounted on the top of the tank* I'm sorry officer, but this isn't a car, it's a military combat vehicle.
 
5
Q: Make Link markdown with whitespace-only content for the 'link' text auto-delete itself

Thomas WardThere was a spam post caught by SmokeDetector containing an invisible link set forth on white space like follows: [ ](hxxp://some-nsfw-image-link-or-spam-link) ... which renders pretty much as a 'space', as is the example here between these parentheses (note that in this case, the invisible li...

 
@ThomasWard OK then please step out of the vodka powered diamond infused madness tank
 
*accidentally hits the "go forward" pedal, and runs over the police car, crushing it* Oops...
ANYWAYS
 
@ThomasWard I just ran through a kind of hilarious scenario in my head involving a future of people with very low expectations, and a time traveler who came back and said that, but it turned out all he could do was divide things really well. And nobody was impressed.
 
Metasmoke has a CI right
 
3:08 AM
Which, in my head, is way better than it is when I type it in chat.
2
 
@JasonC heh
 
@quartata kinda sorta
 
3:29 AM
well...that's what you get for programming in C :D
or Swift on Linux
even though Swift shouldn't ever segfault
I wonder if being a high-rep user on PPCG makes it easier or harder to find a job
 
@QPaysTaxes that'd sound pretty cool
I like the descriptive description for the repo. 'A programming language'.
'A programming language used for programming things'
LOL, you actually took my suggestion
Well, good luck at getting it done sometime before 6-8 weeks are up
 
4:07 AM
ur mom is an exhaustible, non-resettable resource
 
>.>
 
Heh
 
I'm sorry; that's not fun at all.
 
Jeez. Mine just tells bad jokes a lot.
 
C'mon everyone, let's leave the your mum jokes/scary childhoods to chat.urmom.SE
 
4:31 AM
I've got to go to bed now; I need to get up early tomorrow for church. See you later!
 
@NobodyNada o/
 
00:00 - 05:0005:00 - 00:00

« first day (1218 days earlier)      last day (2595 days later) »