And now I can't use SE at all for a little while

@DJMcMayhem wait, really? lol

Yeah...

Too bad there isn't a hide-all button

tryin to join github and the username WELZ is already taken, any ideas what I can use?

I wanted to hide all on just one site, but that hides on every site, so I tried to unhide all

Got halfway through, and now I can't use anything other than chat

owned

@WELZ WellOkThenZ

and it's an inactive user darn it.

@quartata probably dead because datacenter maintenance. too bad my laptop crapped so I can't get in just now to reload them.

@WELZ if you talk to github they can rename teh old one and you can have the new one

however they might say no if that user is active in private repos

@DJMcMayhem Wat.

oh yay i remembered to bring the encrypted drive that has my keys heh ^.^

@Mithrandir You gave me a bad idea. This is all your fault :P

...can you see my IPS account, @DJMcMayhem?

@ThomasWard the chaos football

@Mithrandir No

But I'm not trying to hide IPS from my network (which is easy) I'm trying to hide my network from IPS

Same thing.

No?

My user ID is 31 on IPS.

@DJMcMayhem yea hta'ts how it works

if a site is hidden, it's only visible from itself and it shows only itself

OK, my IPS ID is 3335

Can you find my network profile from there?

Yes, but only if you know how.

Actually, I guess I could check that on my own with incognito mode

@DJMcMayhem you have your flare in the about me section

Also... you still have your flair in the profile text.

Oh. It is hidden. I just always assumed it wasn't because it's not hidden to me

Yeah, I know

Just hide the IPS account, edit the text for that site only, and viola.

and violin too

@DJMcMayhem I can :D

cellos?

@DJMcMayhem however I can currently find the account using the reverse... heading to network profile and it's your top network post

Yeah, I know. :/

I have no idea how to hide that

There is a way to find the network account from a hidden account, but you have to know what to do.

@DJMcMayhem You can't - you'll see that I have that too.

It's not a huge deal to me, but it would be nice to not have that linked everywhere

@Catija don't foget the double basses

@Mithrandir does it require super special mod tools?

[ SmokeDetector | MS ] Mostly dots in answer: RC phase shift oscillator : how to give positive feedback? by jhgj on electronics.SE

@WELZ Nope.

@WELZ a userscript

And o/ all

@SmokeDetector k

@CalvT븃 ohai, and nope

Well I use a userscript to make it easier anyway :P

> Ministry of silly walks

@DJMcMayhem nice location

tyvm

sd k

[ SmokeDetector ] SmokeDetector started in standby mode at rev cd39cae (quartata: Added !!/inqueue command) (running on teward/Solar Flare)

@Mithrandir hint?

!!/standby Henders

Henders/EC2 is switching to standby

[ SmokeDetector ] SmokeDetector started in standby mode at rev cd39cae (quartata: Added !!/inqueue command) (running on Henders/EC2)

pretty much, yes.

teward/Solar Flare received failover signal.

Restart: API quota is 19999.

@WELZ it's a link that I suspect most people don't know is a link.

I assume you just got João's email, @CalvT븃? :D

@Mithrandir got it!

How do you even pronounce a squiggly a?

@Mithrandir I assume you're referring to the accounts tab on the network profile? Normal people can't actually see hidden communities on that

@quartata no, I think he's talking from site to network

@quartata no, other way around

@Riker yes

@Mithrandir ...right. the network profile link. that's what I mean

if you hit accounts from there you can't see the hidden communities

@Mithrandir aye :)

@quartata That's not what they're worried about

No, but if you click it from a hidden account it takes you to the network profile.

oh gotcha nevermind

yeah nothing you can do about that

Finding hidden accounts from the network is harder... unless it's conveniently there in the top posts.

I'm sure you could do it with SEDE

@Mithrandir it only needs to have a score of 5 to get there.

I forgot how much of a pain building Nokogiri is

@quartata Solar Flare's back up

and updated.

noted

I already failed over

Lunar Eclipse has a problem but I can't get into it right now

because it's not responding on SSH

and i'mi operating at 50% capacity because all I have are SSh keys and not all my other creds

[ SmokeDetector | MS ] Few unique characters in answer, mostly dots in answer: Outro vs Rest of Song - Clashing Meaning? The Hills (The Weeknd) by bebbbebeb on musicfans.SE

so working with the company behind it first.

tpu- by DJMcMayhem

@Mithrandir or you can be a mod

grumbles angrily

@WELZ Sure... but, like, I've asked 280+ questions on the SE network. I have a lot of posts with a score of +5... combing through those is a lot of work.

@Mithrandir spywork is a lot of work

@ArtOfCode Indeed. Makes it a lot easier :P

Anyone completed the developer survey yet?

Long isn't it

It's not too bad -- at least the progress bar worked this time

@JakeSymons I hit none of these and they told me I'm ineligible Lmao!

[ SmokeDetector | MS ] Bad NS for domain in body: How the "buy followers" work on social media? by Ashar7 on webapps.SE

You know as much as I hate Node, I have to give it one thing: it's easy to install

@SmokeDetector k

like it has a zip with the linux binaries and it just works

although it is a .tar.xz which makes extra fun to extract

Just got a code chef account :)

no idea what to do though

Any of you guys have code chef?

!!/watch gramcart\.com

@WELZ That pattern looks like it's already caught by Bad ns for domain in body and Bad ns for domain in answer; append -force if you really want to do that.

!!/test gramcart.com

Gemfiles are actually pretty lowkey cool

at least when it comes to something like git_source(:gitlab) { |repo| "https://gitlab.com/#{repo}.git" }

[ SmokeDetector ] SmokeDetector started in standby mode at rev cd39cae (quartata: Added !!/inqueue command) (running on teward/Lunar Eclipse)

LE's back up, @quartata

thanks to the datacenter people for getting me my login creds back >.>

@ArtOfCode So AR automatically gives models an id primary key right

aye

How do I do a has_many from the rails generate ...

you add a :references on the dependent model

ah, so make that first

wait

no

i.e. Post has many Votes, so rails g Vote type:string post:references

OK so the Post comes first with no mention of the dependent

Then generate Votes

ok

aye

you still have to write the has_many line yourself, but it'll do the belongs_to for you

hmm all right

it's not like has_many :votes is difficult to type :)

nah nah

just thinking about how to structure this

I was thinking that Bot would have the name and the info on the actual request to make

@quartata how are you doing it? Starting fresh, copying metasmoke?

and have you got a repo

Then there'd be two dependents BotFeedbackType and BotCommand

Starting fresh, no repo yet because I literally just did rails new

@quartata s/Bot//

@quartata in that case, I recommend adding and initializing gems first

devise, rolify, webpacker, etc

Oh I did that already

not quite just rails new

Although I haven't added webpacker or any of that yet

@quartata okay, lemme know if you want me to do that

either way we're gonna need it for sane JS development

We may not have any JavaScript

Something JF mentioned was just providing an API and then people make webapps for each of their bots

Since I was already doing that for SmokeDetector via MS that's somewhat tempting

@quartata that... sounds like a bad idea

you're gonna expect bot developers to write AWS lambdas and web apps?

The AWS lambda part is just a way they could host their bots but

As I mentioned before the only bot with any exotic dashboard needs is Guttenberg so maybe they don't need that much flexibility :P

I liked the original idea - provide a few communication options (AWS lambda, WS, maybe Google Cloud equivalent), and do it all through that link

Sorry, maybe I wasn't clear. The webapp would be for a dashboard like MS to view posts

It would be pure JS

Fetch posts from our API

@quartata you can do that as part of this as well

well anyways

the gazillion dollar question

Bot requests post -> send post to bot -> receive bot classification and related data and save it to the database

which of these SRP implementations do I want

@ArtOfCode right, that's the current workflow

That's happening regardless

User requests post view -> get view template from database -> get post data from database (the stuff you saved earlier) -> generate view with template and data

i.e. you let bot maintainers create their own view template for the post view

Yes

That was also the original plan

then maintainers don't need to write their own webapps as well, just one view

What JF is saying is to just barf up the why data we have via an API, then people write their own viewer in JS

More power, but not as easy for them

there are good templating engines that would let you do basically the same thing, as part of this rails app, without shoving the extra effort onto bot developers

Speaking of which

I've decided for now to make it so that the bot configuration is purely in the database, and there'll be some sort of /bots/create route that takes a JSON post like the YAML I showed earlier

But

In order to make it fast it has to generate methods and views based on that config (I'm not looking it up each time a post comes through and doing a bunch of branches)

So I was thinking just some sort of after_commit

@quartata s/commit/create/

What if it updates though

you have an after_create and after_update pointing to the same method

after_commit fires after the commit of every transaction, including deletes

oh right

isn't there something like after_commit :method, on: update or something

shrug

anyway

why does it have to do any generation

Is that bug from this morning fixed?

@quartata can you chuck it in a repo at a convenient point so I can have a look?

what 'bug' was going on this morning

@ThomasWard this, nothing was flagging on Graphic design and other sites

Apparently the StackExchange dev team wants to award me for being involved with SmokeDetector. May I get some confirmation before I dump my address into a Google form that I got through email?

@JohnDvorak >>>>>>>

yeah that

oh, nice

thanks

oh, that reminds me

@ArtOfCode you owe me coffee

!!/coffee ThomasWard

@ArtOfCode brews a cup of Cappuccino for @ThomasWard

there ya go

no no no no no not that crap

@ArtOfCode brews a cup of Americano for @ThomasWard

the GOOD STUFF

the stuff we have to IMPORT :P

how's that

why do I owe you coffee anyway

@ArtOfCode brews a cup of Macchiato for @ThomasWard

because reasons :P

yawn

i think i broke Java on this computer lol

or rather, Cisco ASDM did

@ArtOfCode if I remember correctly, you said that I could ask again for privs around this time, is that still the case?

[ SmokeDetector | MS ] Bad keyword in answer: Unable to expand ".zip" into "downloads" (Error - 63 file name too long.) by Kevin Howlett on apple.SE

tpu- by DJMcMayhem

@JakeSymons you do remember correctly... re-read these three things and I'll grant them when I get a chance

@ArtOfCode thanks

Why does it say here to not downvote spam or r/a posts?

or delete them

Most spam posts get down votes anyway though :)

The spam flags will do all that for you.

Flagging as spam will give it a auto downvote and delete it after 6 flags.

@JakeSymons it reduces their visibility, so it doesn't get flagged as quickly

Downvoting spam doesn't really hurt or help us here

We're going to get enough flags very quickly with or without community involvement. Downvoting spam only comes into play if you find it organically and we don't know about it

@Undo That makes sense, should have #scrolleddown it did say :)

@ArtOfCode Only in the case of questions, though?

[ SmokeDetector | MS ] Potentially bad keyword in body: Optimizing a nested for loop append of a large(ish) dictionary in python by Epic Boss on stackoverflow.com

sd why

@NobodyNada [:42161634] Body - Position 4198-4216: 234.76975856506695

fp- by NobodyNada

TL;DR (see my answer there): in most cases, it doesn't matter if you downvote or not.

[ SmokeDetector | MS ] Misleading link: Unable to pip install discord.py in python 3.6 on linux by pop4959 on stackoverflow.com

fp- by Glorfindel

OK, seriously, why haven't we removed that reason yet?

3/57 is a pretty terrible track-record

@DJMcMayhem wow that is bad

Perhaps the matching should be more clever: ignoring common file extensions, ignoring links to imgur and github...

@DJMcMayhem It is a bad track record, but IMO what needs to happen are some relatively easy improvements which eliminate 95%+ of the false positives. If you look through the fp's, almost all of them are just a few types things which we could easily eliminate.

What about having it to only register bad-websites, not just any website on misleading link?

@Makyen Is it worth improving to catch so few tps?

If the TPs it catches wouldn't be caught otherwise, yes.

'username similar to website in answer' has ~50% TP rate, yet (together with 'link at end of answer') is one of our most important catcher of new spam.

The misleading link, when properly fixed, could have a role there too, though it will trigger less often.

[ SmokeDetector | MS ] Phone number detected in title: spectrum phone number 1877-205-0852 Spectrum customer service by Amit Kumar on stackoverflow.com

tpu- by J F

@SmokeDetector why didn't it get "customer service"

@JakeSymons Title - Phone number: 18772050852

!!/test customer service

> Would not be caught as a post, title or username.

Why is customer service not blacklisted?

@DJMcMayhem That is a good question. If the fp rate is low and it doesn't cost a lot of processing, then yes, it's worth it. How many rules do we currently have which rarely catch a tp (and have a low fp rate)? Obviously, if we can't bring down the fp rate, then we should remove it, but the last I checked there were only 1 maybe 2 fp which didn't meet the criteria I listed in the linked message, all of which should be able to be implemented w/o much difficulty.

@JakeSymons because there are many posts which contain those words and are not spam: stackexchange.com/search?q=%22customer+service%22 gives 129,000 hits.

@Glorfindel So far, there's only one tp where the misleading link reason made a difference...

@Glorfindel I checked here where there is a 245/9 tp

[ SmokeDetector | MS ] Blacklisted user: Text fill effect - simulate a block of color passing through text by Code Quit on stackoverflow.com

But I accept what you say, there are a lot of questions

fp- by DJMcMayhem

Metasmoke is weighted towards TPs.

@JakeSymons that doesn't surprise me, but thanks for doing the search.

Predicting a TP/FP rate on metasmoke data alone is bound to fail.

@SmokeDetector f

...as we saw with the 'title ends with comma' reason.

@Mithrandir that didn't really work did it :) 3/10 tp here

@quartata @ArtOfCode about the thing I was talking about earlier: I meant that we would build the website as a single-page app, reducing load on the server. (kinda like how Travis does it)

@NisseEngström I agree removal should be considered. However, IMO, we just don't have the data (or at least not in an easily identifiable organization). The main thing is that the fp rate needs to come down, as the current rate is annoying to people and results in an assumption that the reason is a fp. I agree that we should either improve detection, eliminating the reasons in the linked message (and the follow-on message), or we should remove it. IMO, improving it is the current correct choice.

@JF oh hey just the person I need :)

yes?

@Makyen Agreed.

@JF I'm attempting to install webpack on a rails project. AFAIK I've done the same as I did for another application that it worked on, but this one ain't working. ./bin/webpack-dev-server appears to do a bundle install, then print options and exit

Have you come across that before?

According to the MS code, it runs yarn run webpack-dev-server --progress --color --config config/webpack/development.js. Does running that by itself work for you?

@ArtOfCode I haven’t, but I’ll see what I can do to help.

@JF no, same thing

Specifically:

Missing argument values: config-name, context, entry, module-bind, module-bind-post, module-bind-pre, output-path, output-filename, output-chunk-filename, output-source-map-filename, output-public-path, output-jsonp-function, output-library, output-library-target, records-input-path, records-output-path, records-path, define, target, watch-aggregate-timeout, devtool, resolve-alias, resolve-extensions, resolve-loader-alias, optimize-max-chunks, optimize-min-chunk-size, prefetch, provide, plugin, open-page