« first day (1567 days earlier)      last day (2252 days later) » 

12:00 AM
I'm... gonna push it and see if any MS folks have thoughts. Don't merge it because I need to add some more features.
 
[ metasmoke ] ci/circleci failure on f7db521: Your tests failed on CircleCI
[ metasmoke ] ci/circleci failure on f7db521: Your tests failed on CircleCI
API quota rolled over with 8384 requests remaining. Current quota: 19999.
math: 186
stackoverflow: 135
codegolf: 36
physics: 35
superuser: 34
serverfault: 29
unix: 23
security: 22
askubuntu: 22
english: 17
gis: 12
interpersonal: 12
money: 12
workplace: 11
es.stackoverflow: 11
scifi: 10
bitcoin: 10
electronics: 9
academia: 9
mathoverflow.net: 9
apple: 8
worldbuilding: 7
history: 7
android: 7
arduino: 7
graphicdesign: 6
dba: 6
drupal: 6
gaming: 5
stats: 5
ell: 5
emacs: 5
travel: 5
diy: 5
salesforce: 5
networkengineering: 4
ru.stackoverflow: 4
outdoors: 4
parenting: 4
meta: 4
 
12:28 AM
@thesecretmaster what was the AR
 
[ SmokeDetector | MS ] Few unique characters in body, repeating words in body, repeating words in title: La la la la la la la la by Alisha Rose on law.SE
 
Took ~2 minutes. That's down from the original 10.
 
[ SmokeDetector | MS ] Bad keyword in answer, bad keyword with email in answer: How can I finish marriage ceremony in Skyrim (bug workaround)? by Mercy on gaming.SE
 
@thesecretmaster What are you trying to get? Post counts per reason?
 
sd k
 
12:48 AM
Yeah because if so I'm pretty sure post_reasons has a created_at too
Just do a group by + count
 
I mean it still takes ages, but it's a thing you can do with pure SQL (or AR), without Ruby manipulations too
15.062sec for an SQL query
 
ohh you need a join anyway
for the created_at
well....
 
you don't need a created_at if all you want is all-time counts
SELECT
	r.reason_name,
	COUNT(DISTINCT pr.post_id)
FROM reasons AS r
INNER JOIN posts_reasons AS pr ON r.id = pr.reason_id
GROUP BY r.id
ORDER BY COUNT(DISTINCT pr.post_id) DESC;
 
1:11 AM
!!/notify 11540 chinese True
 
@paper1111 You'll now get pings from me if I report a post on chinese, in room 11540 on chat.stackexchange.com
 
the true is optional
it's default
@ArtOfCode bummer you need the inner join for the reason name, forgot
 
@quartata no you need the join whether you want the name or not... posts <-> reasons is many-many, not one-many where you could just count the reference ID
 
@quartata oh ok
 
@ArtOfCode Any updates on Spike? :D
 
1:15 AM
@Catija no :(
 
Awwww :( Poor Spike. Are you home this weekend or will you be without Squidge another week?
Can you get the parents to mail Squidge to you? :D
 
I probably could, but I'm up there again next weekend so it seems kinda pointless :P
lemme look at the tracking
 
Ah! That's not so bad.
 
Yeah, still nothing new on the tracking
 
:(
 
1:30 AM
Any updates on the stickers BTW?
Also, you two are doing an amazing job
 
@angussidney they're on their way is all I can tell you :)
 
Oh goodie
 
supposedly 5-7 working days from yesterday
 
@ArtOfCode maybe I'm thick but I don't see why we can't count the reason IDs
 
@quartata because there are only 150 or so of those
There's one record in reasons for every unique reason name we've ever created - not for every time it's been used on a post
that's what posts_reasons is
 
1:36 AM
I'm talking about select reason_id, count(reason_id) from posts_reasons group by reason_id
Oh your select was on reasons
I misread
 
Oh, off posts_reasons. Yeah, that would work. You'd want to select reason_id and count(distinct post_id) rather than count(reason_id) for an accurate count, though
@Catija I sent 'em a message. It was meant to be here on the 26th...
 
tpu- by K.Dᴀᴠɪs
 
!!/watch suijuriscourtangels\.com
 
@K.Dᴀᴠɪs You are not a privileged user. Please see the privileges wiki page for information on what privileges are and what is expected of privileged users.
 
@Andy You feeling generous enough to review a 15-second custom mod flag in SO? :)
 
Link?
On mobile, but got some time
 
2:05 AM
It's not to a post. Not sure how to grab a link to a flag
 
Post the flag is on
 
13
Q: Sizeof operator with variable-length array type

rspAccording to cppreference: If the type of expression is a variable-length array type, expression is evaluated and the size of the array it evaluates to is calculated at run time. It means: if the type of expression is a VLA type, then expression is evaluated. For example: #include <st...

 
@K.Dᴀᴠɪs no. Flag it and someone will get to it.
 
Should we watch that site? suijuriscourtangels\.com
 
Yeah, sorry. Two problems there: (1) that button is hidden deep in a not mobile friendly page somewhere, and (2) I don't want an appearance of partially because of the room. It'll get handled.
We need to blitz that queue.
 
2:09 AM
Finally gone through the BI discussion after an hour and a half's reading
Not sure if it's too late to leave a comment or opinion now, but I agree that flagging is our speciality, but not commenting.
 
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, blacklisted website in body: a way to the latest in chronological ord by Vesar Goli on webapps.SE
tpu- by K.Dᴀᴠɪs
 
From now on, I will personally refrain from commenting on sites that I'm not familiar with (like BI, though I don't even have an account there)
 
Yeah I generally comment on SO only. Well, I don't actually - Advanced Flagging does xD
 
As well, I'll re-write my canned comments. Although I have no plan to submit a PR to merge those rewritten ones into Charcoal's, they'll be freely available at my personal GH Pages site, ibug.github.io/ext/doc/StackExchange/AutoComments/…
@K.Dᴀᴠɪs I never have any automated commenting tools ready. I manually click [Add Comment] on every comment I've left so sar.
 
That sounds like a pain
 
I never comment on spam that are not on SO, and I do it manually anyways
 
@K.Dᴀᴠɪs No. I sometimes edit affiliation in directly without leaving a comment when I think it's a decent post.
IMO, commenting is an interaction, or at least an indication that you're open to interact. Automating that doesn't look friendly to me.
 
Well, generally first time users don't know it's automated as they likely have never seen the automated comments before.
 
FWIW, I never respond to batch-sent messages in my social network, despite the difference between a SN and the SE network.
 
If someone commented to me with an auto-comment without ever seeing it, I'd probably not know the difference
 
2:20 AM
I probably missed an exception: The LQP queue
But when you leave a comment, it means you're intended to help the author improve, either the existing post, or their knowledge of submitting further posts.
The exact rules may differ quite a lot across sites (like BI, where the flame arose). It won't benefit much to anyone to "help" recklessly, without a clear mind of what the variations are in place.
Since the discussion was over over a day ago, I'm stopping here. Just sharing a few thoughts of mine.
 
@iBug I agree
 
@K.Dᴀᴠɪs fixed that for you, you should have privileges in chat now when the commit gets pulled in
 
@angussidney ty
 
CI on e9bd247 succeeded. Message contains 'autopull', pulling...
[ SmokeDetector ] SmokeDetector started at rev e9bd247 (angussidney: +K.Davis to CHQ --autopull) (running on teward/Solar Flare)
Restart: API quota is 19416.
 
Try !!/amiprivileged
 
2:27 AM
!!/amiprivileged
 
@K.Dᴀᴠɪs ✓ You are a privileged user.
 
\o/
!!/watch suijuriscourtangels\.com
 
@K.Dᴀᴠɪs You don't have code privileges, but I've created PR#1670 for you.
Merged SmokeDetector #1670.
CI on 24d6498 succeeded. Message contains 'autopull', pulling...
Restart: API quota is 19407.
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, blacklisted website in body: ou will likewise soon find that by Seg bjkj on apple.SE
tpu- by K.Dᴀᴠɪs
 
@iBug Fast reader ;)
 
@Undo Not fast. I stopped where one page of transcript spans across 3 hrs
 
3:08 AM
@Undo planning to run primarymonitor if SO has an election?
and if so want a CNAME off charcoal-se.org like the last one?
 
[ SmokeDetector | MS ] Manually reported answer: How to read arrays from user in java? by howard on stackoverflow.com
naa- by ArtOfCode
tpu- by Rob
 
3:23 AM
@ArtOfCode Posts counts per reason in the last n months
 
So yeah, that query I posted with an extra WHERE
SELECT
	r.reason_name,
	COUNT(DISTINCT pr.post_id)
FROM reasons AS r
INNER JOIN posts_reasons AS pr ON r.id = pr.reason_id
WHERE pr.created_at >= '2017-01-01'
GROUP BY r.id
ORDER BY COUNT(DISTINCT pr.post_id) DESC;
 
Ah, you're counting post_reasons rather than posts. That's what I was missing.
Actually, no. I remember why I didn't do that -- I need it to be Posts per reason per site created after <date>
 
tpu- by K.Dᴀᴠɪs
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, bad keyword in title, blacklisted website in body, pattern-matching website in body: Trilixton muscle building supplement. by NoeKones on drupal.SE
tpu- by K.Dᴀᴠɪs
 
@thesecretmaster You can count posts, that also works, it just takes longer because it's doing double the joins
that query gives you posts per reason created after a specified date
grouping on sites as well, yeah, requires posts
 
tpu by gparyani on La la la la la la la la [MS]
 
3:38 AM
Which, I think, takes (significantly?) longer.
 
@thesecretmaster potentially. Let me get the latest dump onto this machine, then I'll do some queries. The date restriction helps a lot, might be possible to nest it lower in the query to optimize further
 
I'm going to check again after I migrate my dev env to my desktop from my horrible laptop -- by the way, metasmoke dev froze then killed my computer earlier today.
Probobly because mysql
 
uh
that sounds like a bigger problem than mysql
 
I promise. I had a browser open and terminal and nothing else.
It's a really sad laptop
 
Even long-running queries shouldn't be locking you up unless you're very restricted on power, but IIRC you're on a mac with 32GB
 
3:43 AM
No, I'm on linux with 32gb and a nice CPU and a GPU, and I have a 2011 mac with 4gb.
 
fp- by K.Dᴀᴠɪs
 
It killed the mac. It'd be fine on my linux box.
 
@K.Dᴀᴠɪs @gparyani Re: Why did steam locomotives not transfer power by cogwheels? [ MS ]: While I initially agreed with you that it's spam, upon further consideration, I have a hard time believing this is anything other than a naive user attempting to provide a "helpful" link to an example. I might go with tp, but I wouldn't flag it as spam (tpu).
 
[ SmokeDetector | MS ] Offensive answer detected: Do solids have translational energy? by bitch on physics.SE
tpu- by K.Dᴀᴠɪs
 
@Makyen agree on naa for that one
 
3:49 AM
@Makyen Ehhhhhh... Idk, it's like a question asking "How does a boat work" and you get the answer "Take a cruise on the wonderful xyz ship today! Visit here to book your date"
Maybe less dramatic, but he did link to a company's website that exists to sell a service
...and the fact he has no other network profiles, never made a previous post
 
I don't see how a question about how steam engines work could lead a misguided user to post something about a random non-piston railway
 
@K.Dᴀᴠɪs I agree that it's technically spam (unsolicited promotion). I just feel the user should be given the benefit of the doubt, rather than be slapped with the consequences of a spam-delete.
 
@ArtOfCode Might be good PR
 
@Makyen Even if that's the case, SD still did the correct thing by catching it since it's "technically spam" - regardless of what happens to OP; thus the feedback to SD imo should have been left at tpu... technically spam is still spam.
 
(main reason is to obscure my domain, though)
 
4:01 AM
@Undo yeah, either way. Lemme know the address and I'll throw a record up
 
@angussidney The question is about steam locomotives not using cogwheels. While I haven't gone to the linked site, the name of the place, "Pikes Peak Cog Railway", implies it's a counter example to the question OP's assumption that steam locomotives didn't use cogwheels.
 
FYI the site is perfectly reasonably (as can be expected) safe, I've opened it
 
@K.Dᴀᴠɪs The only part of that I disagree with is tpu instead of tp. Which to use (tp vs. naa) depends on how nuanced we want the feeback to be. I'm fine with either one, but more concerned with the spam flags.
@angussidney Random vagaries of humanity. I've certainly known people that might post that.
 
Fallback option: ask a mod?
 
fwiw I don't think we'd be having this conversation if it weren't for recent events
 
4:08 AM
Looks like a confused user to me
Assuming you're talking about engineering.stackexchange.com/a/19701/84
I'd decline a spam flag if I couldn't find anything on the profile
 
@Undo that.... or a witty spammer who knows how to bypass the rules and get away with it.
 
@K.Dᴀᴠɪs I expect that I certainly would have at least posted my initial message, even without recent events.
 
I've been on that train. It's kinda neat.
 
@K.Dᴀᴠɪs That's... possible. Any way you cut it, it's at least NAA. Thus, it will be deleted. If what you posit it true, it'll just live for a while longer than other spam.
 
I don't really care what happens to it, just don't nuke it. Give a mod a chance to look at it.
3
 
4:16 AM
anywaaaays... I think autoflagging flag numbers should be based off how active a site is. I am curious to see a report of 1 v 5 flags on a site like SO vs a site with much less activity.
 
Grrrr.... webpacker....
 
@K.Dᴀᴠɪs I've thought of a lot of clever ideas like that... but ultimately, we can afford to run 5 flags everywhere with basically zero projected risk.
Usually we do clever things to exclude false positives, or reduce their impact. This is a really weird case for us because there are no false positives over about 300 weight.
 
Given the debacle we had yesterday, I'd suggest that we hold off on that meta post on autoflagging for a week or two
 
Agree
We probably will anyway while the data from the experiment flows in
There's definitely a balance to be struck between waiting for people to settle... and not giving them an opportunity to being unsettled again.
Once that meta post goes out, gets featured, and receives good reception - as it will, 95% guarantee - we're in a far better position with regard to the 'community agency' arguments.
We did this a year ago. It worked great, aside from the mSO kerfuffle.
And remember, a majority of people just aren't interested in what we do beyond a "that's cool" level. We hear the vocal minority, and the vocal minority is almost never the supporters. The world isn't against us, even though it can feel that way.
 
@thesecretmaster query:
SELECT
  r.reason_name,
  s.site_name,
  COUNT(DISTINCT pr.post_id)
FROM reasons AS r
INNER JOIN posts_reasons AS pr ON r.id = pr.reason_id
INNER JOIN posts AS p ON pr.post_id = p.id
INNER JOIN sites AS s ON p.site_id = s.id
WHERE p.created_at >= '2017-01-01'
GROUP BY r.id, p.site_id
ORDER BY p.site_id ASC, COUNT(DISTINCT pr.post_id) DESC;
 
4:26 AM
@Undo A vast majority of people don't know Charcoal exists, AFAIK :)
 
@Art... I don't want to outright rudely accuse your programmer dog of eating Squidge but...
 
@thesecretmaster Indeed. I like it that way ;)
 
@Catija Weird, that didn't ping me
 
The dots
 
Too many dots, I think.
 
4:27 AM
@Catija. This won't work either
 
oh
 
@Undo Yeah it did
 
o.O
Special-cased?
 
Thanks for the query
 
4:27 AM
@Catija.. This?
 
@Undo no. That's why I said "too many". Single punctuation marks work fine.
 
[ SmokeDetector | MS ] URL in title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +2 more: www.ifirmationeyeserumblog.com/tryvix-italy/ by jicwongd on english.SE
tpu- by K.Dᴀᴠɪs
 
> We'd like to cast more flags on spam automatically. What do you think?
I think we need to work on that title
 
Silly @Undo! That's not how chat works! :P
 
That one worked too
Does two bangs work?
 
4:29 AM
Nope, @Undo!! I don't think so.
 
yep
that worked
Maybe it's the shortening @Catij!!
 
that worked fine.
 
well that's odd
 
What about this @Undo?
 
Anyone got webpacker troubleshooting tips?
 
4:30 AM
You can tell when it works because the hover highlighting points at their previous message.
 
Worked
 
maybe stop using webpacker?
 
@thesecretmaster Got a traceback?
 
suppose that's not great advice but
 
A whole bunch of stuff. Error line: Error: Chunk.entrypoints: Use Chunks.groupsIterable and filter by instanceof Entrypoint instead
 
4:32 AM
Is your Ruby/Rails version right?
Never seen that one
 
2.5.0
rails 5.2
 
Anything more than that one line?
 
what version of node
 
Yeah, a whole traceback from there. And there were a couple other errors that I "fixed" before getting this one.
 
Did it ever work on this setup?
 
4:33 AM
No, setting up a dev env on my desktop now
Before, I was using my laptop
 
you know what screw these web technologies that think it's ok to break compatibility like that and also create ecosystems that encourage you to use poorly supported plugins
I joke about it but now it's starting to piss me off
node please
 
Will it run without webpacker? Let's find out!
Nope
 
4:39 AM
[ SmokeDetector | MS ] Bad keyword in answer, blacklisted website in answer: How do I treat my friends to something without it being awkward? by Minnie Christian on interpersonal.SE
tpu- by K.Dᴀᴠɪs
 
What's webpacker responsible for? CSS+JS only?
How hard would it be to migrate away?
 
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, blacklisted website in body, link at end of body, pattern-matching website in body, +2 more: Soyeux Cream Soyeux Cream by soinexna12 on askubuntu.com
tpu- by K.Dᴀᴠɪs
 
@Undo Looks like just JS to me, but I have no idea
 
Sounds right
 
The asset pipeline is supposed to do what webpacker does anyways?
 
4:41 AM
MS doesn't have that much JS, but I know we moved to webpacker to get ES6 (@JF's work)
 
I don't know enough about JS to really be able to say anything about this though. I'd like to migrate off :)
 
I'm sure we could do this in pure Capistrano
 
@Undo yep, webpacker does JS. No ES6 or modules/requires/imports without it
 
Call Babel directly?
it's pretty easy
 
Just as much of a pain as webpack
believe me I've looked at this stuff
Right now we have our JS nicely split across multiple files, one per controller, which all get required from a single central JS file.
 
4:46 AM
How does one destroy all of webpacker and try a reinstall?
 
I have a Cakefile I'm using to build my coffeescript currently
 
It's good that we're able to do that without having to write the transpilation etc ourselves
@quartata it's not coffeescript
 
I just do a glob and run each file through uglify and what not. It's very easy
@ArtOfCode no, but Capistrano has similar capabilities to cake
 
How hard would it be to include only the files for the relevant controller in the file that gets delivered, to decrease page load times?
 
Probably not that hard? Grab the controller name in layout.html.erb and create a script tag based on that?
 
4:49 AM
Rails already does this when building views if I'm not mistaken
 
@quartata not with webpacker
 
Rails works nice with coffee, but less nice with plain JS
 
@ArtOfCode right.
 
@angussidney Page load time isn't driven by asset time, last I checked.
 
Wait... isn't sprockets supposed to do the requiring thing that webpack is doing now?
 
4:50 AM
@thesecretmaster that wouldn't work - the script tag isn't the point at which we choose what to require, that's in the required file
 
Babel does ES6 imports
If I'm not mistaken
 
@Undo given the time the SQL queries take it woudln't make much of a difference
 
@thesecretmaster in plain Rails, yes, not with webpacker
pretty sure we have webpacker on SCOT
or maybe that was another project
 
I don't think so?
 
yeah, that was another project
 
4:51 AM
@ArtOfCode So why not take away webpacker and use sprockets for the imports and such?
 
7 mins ago, by ArtOfCode
@Undo yep, webpacker does JS. No ES6 or modules/requires/imports without it
 
The thing about webpacker is that we use bootstrap which is already minified on the CSS fromt, and using it just for Babel is massive bloat
 
It's not just babel, though, the whole yarn/webpack ecosystem lets us pull stuff from npm, too
 
@Makyen Thanks.
 
i.e. we can require npm packages straight in browser code
 
4:53 AM
Babel takes in a file, spits out a file babel input.js --out-file output.js
@ArtOfCode I know there's a Babel plugin that handles inlining imports
let's face it we need @JF
 
If there is an easier solution, I'm all for it. I'd rather not be maintaining code that manages the requires and transpilation, though - leaving that to people who know more about it than we do is a good place to be in.
 

« first day (1567 days earlier)      last day (2252 days later) »