Tavern on the Meta - 2018-08-03

« first day (2921 days earlier) ← previous day next day → last day (2083 days later) »

Journeyman Geek

12:20 AM

@Catija thanks. But I should also do a little more than saying I hate this on chat :p

Though since folks are aware no need :p

2 hours later…

Somewhat Memorable Name

2:13 AM

Been a while on Meta stalking...

Math.SE has started their election, now on nomination phase with 5 candidates for 2 positions.

Journeyman Geek

I read that as meta.se and went whaaaaaaaaaat?

Sonic the Inclusive Hedgehog

2:25 AM

I also noticed that only the top 20 nominees advance to the primaries, where it's at 30 for other sites.

I think they have a higher minimum rep to run, too.

Sonic the Inclusive Hedgehog

@Catija Yep, they require 1,001+ rep.

@Catija Speaking of the new Code of Conduct and comment flags, maybe it's time to retire the 4-day aging period for "rude or abusive" comment flags? Other comment flags don't age.

Somewhat Memorable Name

2:42 AM

Huh... there are still only 4 answers out of 5 candidates...

Sonic the Inclusive Hedgehog

@SomewhatMemorableName One was nominated within the last hour...give them some time.

Somewhat Memorable Name

Except that one has posted the answer almost at the same of their nomination :)

1 hour later…

Sonic the Inclusive Hedgehog

4:15 AM

@Shog9 Any progress on the vote buttons appearing orange in IE?

...i don't really care that much tbh...

Sonic the Inclusive Hedgehog

@Shog9 Okay...how about this?

Jul 28 at 22:34, by Sonic the Inclusive Hedgehog

@Shog9 Can you please investigate this?

9

Q: Revision history says these two questions are deleted, but they still exist

While playing around with some SEDE queries, I came across the following two odd questions (both from the same user): 758706: Insert matching parentheses xcode? (revision history) 1101705: Input Manager in Cocoa? (revision history) Both of these questions, according to the revision history, w...

support specific-question deleted-questions revisions-list

Revision history for such things is often inaccurate for older posts. There are also examples of posts being undeleted with no delete entry

Sonic the Inclusive Hedgehog

@Shog9 Older Roomba deletions, for example?

Finally, can you please answer this?

0

Q: What sites have the code formatting filter enabled?

I know that the code formatting filter ("your post appears to contain code that is not properly formatted as code") is enabled on Stack Overflow. However, are there any other sites where this filter is enabled? For example, Code Review or Code Golf? Can the code formatting filter be enabled on p...

support markdown code-formatting

Entirely possible someone just ran a update Posts set DeletionDate=null where... at some point. Who knows

1 hour later…

Sonic the Inclusive Hedgehog

5:30 AM

cv-pls meta.stackexchange.com/questions/313557/…

Journeyman Geek

cv-plz meta.stackexchange.com/questions/313558/…

I don't want to bust careers/blogs/advertising's chops again but... this feels like someone who could be a SO Careers/Jobs customer who dosen't seem to know they exist :(

→ 14 messages moved to Chimney

6:14 AM

morning

Sonic the Inclusive Hedgehog

6:27 AM

cv-pls meta.stackexchange.com/questions/313565/…

Somewhat Memorable Name

6:43 AM

Friday.... so lazy to wooooooork

Journeyman Geek

Hah. I work weekend nights this week

7:00 AM

Yeah, I'm lazy too. I start work at 09:00 ... oh ... that is now ...

7:16 AM

Hey @rene

I have a problem with SEDE, mind taking a look?

@Magisch I have 30 minutes, max.

COALESCE((SELECT
SUM(CASE WHEN v2.VoteTypeId = 1 THEN 2 ELSE 0 END)
FROM Votes AS v2
INNER JOIN Posts AS p2 ON v2.PostId = p2.Id
INNER JOIN Posts AS p3 ON p2.Id = p3.AcceptedAnswerId
WHERE p3.OwnerUserId = ##UserId##),0) AS ReputationFromAcceptingAnswers

This times out for me, although it shouldn't be many posts. Why?

do a where votetypeid = 1 and get rid of the case when?

also p3.posttypeid = 1 right?

asnd p2.posttypeid = 2, right?

also: we do have a SEDE chatroom now as well: chat.meta.stackexchange.com/rooms/1223/data-explorer-sede ;)

Ah

good idea

limit the index scans

yes. Keep in mind that the Posts table is a VIEW over the table postswithdeleted. Being explicit about which posttypes your'e looking for cuts that resultset in half if not more.

Also the votes table is huge. always try to limit it with a where

7:28 AM

COALESCE((SELECT
SUM(2)
FROM Votes AS v2
INNER JOIN Posts AS p2 ON v2.PostId = p2.Id
INNER JOIN Posts AS p3 ON p2.Id = p3.AcceptedAnswerId
WHERE p3.OwnerUserId = ##UserId##
AND p3.PostTypeId = 1
AND p2.PostTypeId = 2
AND v2.VoteTypeId = 1),0) AS ReputationFromAcceptingAnswers

Still times out.

Does it make a difference if I originally select from Posts P3 and INNER JOIN on the other 2 or is that the same thing?

yeah, get rid of p2 and do INNER JOIN Posts AS p3 on v2.Postid = p3.AcceptedAnswerId

but I assume you run this on the column of a resultset?

Yes this is part of a result set

I'm working on this: meta.stackexchange.com/questions/313561/… via the approach of first calculating the reputation a user has earned a day sans cap and then selecting from that resultset as subquery

This is the rep gained for questions you asked and accepted an answer, right?

Yes

the +2 you get from accepting. I already have rep from post up/downvotes, accepts recieved, bounties recieved, suggested edits accepted and bounties offered

select count(*) * 2 from posts where posttypeid =1 and OwnerUserId = ##UserId## and acceptedanswerid is not null

7:39 AM

Thats actually a lot better

it is shorter. If it still times out we didn't gain a thing

I'll try

Even without all of that the result set takes 23s to execute so I'm straddling the line a lil

23s is within the 120s you're allowed

COALESCE((SELECT
SUM(v4.BountyAmount * -1)
FROM Votes AS v4
WHERE v4.VoteTypeId = 8
AND v4.UserId = ##UserId##
AND v4.CreationDate = v.CreationDate
),0) AS ReputationOfferedAsBounty

Also causes it to time out strangely

I'd think that I narrowed Votes down enough but apparently not

v.CreationDate is what the main result set is grouped by

there are no indexes on fields like creationdate so once you end-up doing that you can only pray.

7:45 AM

That stinks

So it might not even be possible is what you mean

Unless on a smaller site

Wait

@Magisch depends. Maybe for a user with little to no posts.

The bounty stuff doesn't need to be grouped per day anyways, it's unaffected by the rep cap

also: meta.stackexchange.com/questions/256115/…

I can do that in the superceding query anyways

And avoid the creationdate alltogether

@Magisch more relevant: meta.stackexchange.com/questions/268340/…

Okay, now I really need to get that presentation ready ...

8:30 AM

Took a bash at it using SEDE: meta.stackexchange.com/a/313570

Alternate route is of course the API reputation history

8:43 AM

Heat seems to be getting to the bosses too

Somewhat Memorable Name

9:26 AM

Very unimportant remark: my office modem is broken and there's no internet since 3 days ago, relying on Mi-Fi instead...

and the Mi-Fi quota.... was gone since yesterday, thus no internet...

(hmm, how many times I have shared a similar case...?)

Journeyman Geek

ow

(my workplace has no internet. We have a work dongle on your laptops... and my specific phone company has terrible reception... only in my specific office. its great outside ._.)

2 hours later…

Somewhat Memorable Name

11:30 AM

MartialArts.SE needs YOU!... or someone else to be appointed for pro-tempore...

I wonder if anyone here is active on that site

2 hours later…

1:36 PM

Hey @TimPost, I didn't saw the invoice when I told you that I've to pay ~$14 as customs duty. I saw it today and I realize it is not the customs duty. The customs duty is INR 269, which is around $4. DHL charge a fixed price of INR 500 (~$8) and that's why the charge is too high.

user image

Last year, I got the hacktoberfest swag via postal and I don't have to pay anything. Yesterday I got another swag from Antihack, a Singapore based company, via postal and also I didn't paid anything.

user image

status-feature-request

Can you consider sending swag via postal, since they are government service, they won't, at least, make huge profit

@OptimusPrime Did you anonimize the generic stock photo girl? :D

yeah

:p

Nice touch ;)

not everything. just iris and lips

I wonder what crime she committed ;) Even the most hardened criminals over here don't get their lips censored in news reports ;)

1:53 PM

One Jump Man

2:29 PM

@OptimusPrime We can't. 9/10 times that we use local postal services, it never gets there.

But, we should be charged the shipping fee, not you. Odd. I'll find out about it

3:14 PM

Why does it seem like more effort is required to post a comment than is required of people wanting expert help? — Plutonix 2 days ago

^^^ sounds like an accurate summary of the recent CoC / comments flagging activities

3:29 PM

Frustration on an unrelated site: If enabled, your network may see this change with what looks like a switch that says Yes. You can't switch it to No.

Image: i.stack.imgur.com/JQOdK.png

@TimPost I think they already charged the shipping fee from you. The 500 is charged as "Service charge for customs clearance", I guess

They're making over-profit. I don't know why you said it didn't reach the destination 9/10 times. I received 2/3 items from US via postal service and their delivery is fast enough.

bye everybody. Taking some off-work time for the next two weeks, so don't panic if I won't be here to annoy you all

@OptimusPrime It could be that recently things improved, or just the area that you live, but we've had incredibly poor luck getting stuff to people in general much less on time using parcel post there. Half of that problem is probably our (as in the USPS) postal service.

I'm trying to prove that mathematics forms a meta-abelian group, which would finally confirm my suspicions that algebreic geometry and geometric algebra are the same thing.

@Derpy take care! Bring some ponies!

3:44 PM

@ShadowWizard BTW, I am also trying to move the Sugarcube room to the other chat server (details in the room) so If you want to drop a message there to avoid room freeze I would be thankful, but there is no need to overwork it. In the worst scenario, poor @TimPost will have to reopen the room in two weeks from now.

@rene thanks. Take care

4:32 PM

Just wondering has there been any updates on the new site theme roll-out? Was looking forward to trying it on a beta site I visit but it appears to be delayed.

there was a post that outlined the process, it looked like a 2 month repeating thing that rolled it out to n sites at a time.

iirc

@JoeW it looks like they didn't make Late July: Post next 10-20 sites for comment nor did they role out all of those sites from the 1st batch

also lot's of status review / status planned tags in the answers. I guess that sanked the plannig.

4:54 PM

@Magisch something I didn't offer because I hardly ever use it but you could create temp tables and indexes on them. This can work as there is no space limit (at least I've never hit it) and once the temp tables are filled and the right index added the queries over them might run fast enough to compensate for the time needed to create all that #temp stuff.

@rene I ended up getting the thing to work by drawing the more expensive groupings by votes.CreationDate out of the inner query

Takes ~32s for Jon Skeet. Interestingly it doesn't seem to hugely scale with # of reputation events as it takes ~24s for me.

I made a post on codereview.se though if you want to chip in

checking

codereview.stackexchange.com/questions/200880/… link for the lazy

@rene Thanks, I was guessing that is what happened but was wondering/hoping something had been put out about a new timeline.

Maybe Joe Friend will hop in during Happy Hour and will give an informal update.

Sonic the Inclusive Hedgehog

5:12 PM

@TimPost One time a letter we sent to India via USPS didn't arrive until a month later, even though we paid $32 for the Priority Mail service. (cc @OptimusPrime)

@rene He hops in here if you invite him

Joe seems to be everyone's friend...

@SonictheInclusiveHedgehog I added the tag you suggested in your comment. Thanks for that!

Hey, it's Juan

@Magisch there is a bug in how you calculate suggested edits. If there is no vote on a specific date, suggested edits will not be counted for that date

@rene How do you mean?

Suggested edits only count when they're accepted, so only when the last approve vote is cast, no?

5:21 PM

@Magisch yes, 'tis I

I spot checked this with 2 suggested edits from my rep history and it worked

@JuanM seeing more of you in chat recently

It's nice to stretch my legs out, you know? ;)

As you can see, I am a cactus

So not really :p

@Magisch but you link on dates in the votes table for your own posts. Suggested edits can happen anytime by you, not only on days when someone votes on your posts

Right

So if a edit happens on a date where there's no post vote, it won't count?

Not sure how to fix that without destroying the rep cap association

Maybe with another subquery in the topmost query excluding suggested edits on dates a post vote happened

5:24 PM

If there is no vote on a date that you made a suggestededit the suggested edit rep will not be counted, right.

That would of course fail for dates where the suggested edit but no post rep exceeded 200

>_>

I can still suggest an improvement for that part but I didn't anticipate to fix bugs as well.

:(

I didn't think of that tbh

are partial answers OK on CR?

think they are

πάντα ῥεῖ

5:26 PM

es (pun intended ;-) )

Or do I need to CR the whole thing?

How is your overall impression of the query?

I have the feeling you do too much work in the projection phase of the query.

πάντα ῥεῖ

We usually start reviews for c++ with Don't use using namespace std;

In other words, you crammed a lot of stuff in the columns of the select while you left an opportunity to reduce the number of rows to be considered in the first place.

And that needs then be solved by the group by

5:29 PM

You mean I could pull bounty rep earned and accepted answers rep out of the inner query?

πάντα ῥεῖ

That's one of the cool things at CR, you can iteratively add more stuff to your answer as soon you recognize them. You've got all time of the world.

I spent the time writing that query mainly to improve myself at TSQL. So expert feedback is always welcome

@Magisch On first impression that could be things that I would look into

do know I'm not used to doing reviews. specially not for t-sql so I rather test things first before I offer any solid advice

I did consider that but was like "It already does the full scan on votes might as well use the results in this go" but you're right of course

Less rows to consider the better

so in the end I might still end-up with your query

5:31 PM

I should also have limited the allowed votetypeIds more

5:47 PM

@Magisch - If you want rep from date, pull it from the svg in the user profile.

<1s for any user, any exchange

@TravisJ I know about that, but OP was asking for a method to do it programatically. If you do this with a scraper unless you're using it for private use only you're violating the network TOS

It isn't a violation to scrape one page.

If you ran it for every user, I agree, but for one user per submit by manually entering a userid? No way.

I assume if OP wants to do it automatically they're looking for more then one page

Hm, maybe to measure inflation

Fair enough

it was also a fun exercise to do in SEDE

5:53 PM

So can you give me a box and whisker plot of user reputation by day? :P

(only half joking)

Not particularly no

Full table scans of votes by day are extremely expensive in SEDE

493058345ms
5489123ms
13132564ms
2165498ms

That's probably why they batch update at 3am

to do that you'd have to take your chance with the api / reputation private route

And hope you don't have more then 1.000.000 reputation events (you dont)

You had me at api

6:17 PM

4

Q: New "stock theme" roll out

We are rolling out the new "stock theme" to the Stack Exchange network. You all are the lucky first site to get the new theme. Your site should get updated around 3pm eastern today (8/3/18). What new theme? If you're like, "What the heck are you talking about?", then you should read the Meta St...

discussion user-interface design

πάντα ῥεῖ

He he, heading for holidays with mum and daughter:

user image

I am planning to stick that prominently in the car :3

@Magisch wrote an answer

Now I'm going to make dinner

πάντα ῥεῖ

Good appetite

So, when are we just going to disable comments entirely and provide some rigid radio-selection for anonymous feedback that doesn't fit into our current flagging system?

As soon as I figure out which shoe goes on which foot. Priorities ....

πάντα ῥεῖ

6:29 PM

@TimPost Thumb's right means left hand

Very similar.

@TimPost All kidding aside, is that potentially where this whole mess is heading?

Too many naysayers to solve problems anymore.

Asking a question has become a gauntlet of you didn't do it right versus let me help you solve that.

Most answer metrics are trending down and have been for some time.

@πάνταῥεῖ They are going to eye roll as soon as they see it

πάντα ῥεῖ

But I can point them to my third statement :3

What about uneven eye rolling?

πάντα ῥεῖ

6:36 PM

I can point them to my third statement

@canon Not quite. We need to make the most common feedback be system-handled events. E.g. you click "This post lacks needed information" - a dialog comes up and says "What, specifically?" you put it in, and it gets delivered to the OP as a system message that they see, with an edit link. Something (sort of) like that, so we don't depend on free-form text unless it's necessary.

@πάνταῥεῖ It only addresses even eye rolling.

πάντα ῥεῖ

Any

even is underlined

:D

It helps two things. 1 - the feedback is coming from a machine, less likely to be taken personally, and 2 - we can use vetted, effective UX copy to deliver feedback consistently.

But we're still talking about it. Internet comments suck universally, our's suck way less than most, but marginally better than youtube isn't great, so there's still work to do.

6:38 PM

@TimPost I wonder how far that goes. Tag-specific stuff too, e.g.: SQL injection warning?

πάντα ῥεῖ

@TravisJ Because that's a screenshot of the libre office doc. There won't be any underlining in the print version ;-)

@canon It could go to stuff defined in a tag wiki, yes. We haven't ruled that out.

@TimPost like you've never snapped a keyboard over your knee from a cryptic compiler message...

@TimPost 1: At the point a post needs feedback, it's too late. Not only will users rarely respond to feedback after creating a post, the community will take the post in its first version and action it far before they can edit it. Canned feedback in any form is least likely to succeed. 2: Consistent UX helps, but it will not overcome the aforementioned issue in #1...

@TravisJ Not necessarily, if we throttle visibility based on initial reaction, it could in fact be more helpful

6:40 PM

There are too many users looking to shut down questions versus answer them. At Stack Overflow the pool of answerers is ridiculously tiny compared to the rest of the site, and there are so many users who would rather not answer or see someone else answer that the main focus has become shutting down questions.

πάντα ῥεῖ

Seems natural for a FAQ model.

@TimPost If that's anonymous then that's something I've wanted for a long time

@TimPost In many instances, there is nothing that can be done by the question asker that will make the users who would shut down a borderline question change their minds.

That's fair, and true to an extent, but secondary problem to solve.

Anonymous canned comments would be really nice

2

6:42 PM

I disagree that it is secondary.

The closure system hasn't been analyzed for half a decade.

Guiding how people optimize for what to do is another problem we have to solve. We may need to solve that first, but it's separate (perhaps not secondary, but separate)

The closure system is gonna get an overhaul, too.

Well, at least duplicates will.

The least needed category to change.

I'd like to see some data for how often people manage to dig themselves out of downvote hell

after acting on advice

@TimPost There ought to be a way to opt into getting notified if a post you closed gets major edits

mhmm

even downvotes

6:44 PM

@canon almost never, but I haven't run the numbers recently.

@TimPost that's what I expected.

I really want to use my reopen votes, and there are a bunch of posts where judging if I should or not would be really easy, but unless I keep a spreadsheet and check back on my own, I'll never see them

Downvotes are a response to the defined "use" of posts as it relates to topicality. By defining a large amount of useful posts as off topic through vague interpretation, many questions end up being downvoted, ignored, or closed, which would have otherwise been useful.

@canon We almost did that, but it would be notification hell for the majority of people that actively vote.

@TimPost Make it opt in then

6:44 PM

@TimPost Good.

I'll take reputation hell

@TimPost well, not if you opt-in on a per-post basis and aggregate notifications per day.

so often I see a post where I think "This could be good if OP just..."

And it always ends there

I've often downvoted, letting the poster know to ping me if they make changes; so, I can reverse my vote.

Because I can't be bothered to check back

6:45 PM

@canon Per post is an interesting idea. That would make it manageable. Could tie that to the currently semi-useless "favorite" functionality.

But we'd need to extend that to answers. Might as well just call it "watch" and do that.

@Magisch That is because "guessing" is now interpreted as negative. As an expert in the field, often such an informed guess would be seen as a welcome suggestion, even if incorrect.

I think commenting is in a good place with the recent work. If only you guys could come up with a blog post that encourages actually letting users answer.

imean.. what's stopping people from answering

Users enforcing their version of question control by downvoting answers.

so what, it's 2 rep, you get 10 for an upvote and 15 for an accept

@TravisJ What if, to close a duplicate, you had to answer it with a link, and a brief synopsis of how it relates to the question as stated? :)

6:50 PM

It dissuades answers @KevinB. Answers are in bad shape right now because no one answers anymore.

i wouldn't dupe close anywhere near as often

@TimPost I'm fine with that, although I don't see it as solving a problem which is as high priority as others.

@TimPost would that give reputation?

@TimPost Ugh. No. That's adding more steps to a process that already causes complaints and it spits out system generated messages now.

If so, it'd actually work out as incentivizing closing dupes over answering them as well

6:51 PM

@TravisJ I do that! :/

@TimPost Oh, damn... more work and no rep. Sign me up!

@canon Yeah, sorry, but that behavior has been damaging the site.

We can't even find dupes efficiently as power users...

Noone answers the good questions anymore. the super easy dupes still get quick answers every day

with or without downvotes

@TravisJ from my position, answering crap questions encourages crap questions. You disagree?

What I'd love to do, and I'm not sure it's even possible, is somehow classify questions as hard when appropriate.

6:53 PM

@canon Yes, and so does years of evidence from Shog and others.

But 'just needing rare knowledge' != hard.

What is hard in the context of a programming question?

"A lot of work?"

.... I have a feeling tim has me on ignore

@TimPost what, knowing some arcane, undocumented command line switch isn't hard? :P

What happened to Project Reduplication? I never saw results

It would have to be peer-evaluated. "This could be done, but 'how' is going to be super tricky" I'd almost like a button to push that sends that signal for something I find really interesting but can't manage to answer properly. I know we have bounties, but I'd love a way to mark it as 'difficult' so someone looking for difficult questions in , say , c++ could find it easily.

6:57 PM

I have a hard time imagining what a "difficult" question would be, other than a question that is simply unclear.

@KevinB probably something that has an unintuitive/undocumented behavior... or a behavior which requires a deeper understanding of the system, engine, compiler, etc than might normally be required of a developer.

It'd be tough to qualify, for sure.

that's just a different skill set. Though, i'd agree said skill set is more rare.

@KevinB Is that is because you interpret problem solving problems as off topic?

7:13 PM

no?

Sorry, I read your comment out of context.

problem solving problems are easy. the "hard" ones are simply ones that haven't been broken down into smaller pieces.

I missed the scare quote reference.

Why should problems be broken down into small pieces?

i have no idea why i'm quoting random words

We have 30,000 characters to use before running out of space.

7:16 PM

Because often the problem is a duplicate, hidden behind a lot of words/variables

Just as often, the composite of broken down problems creates its own separate issue.

7:34 PM

Is minitech still around? What's his handle now?

@canon Yes. His regular name.

Sonic the Inclusive Hedgehog

@canon Ry-

@SonictheInclusiveHedgehog ah

@Magisch I ran the numbers on this a while back; it'd be bad for some people, not so bad for most... But noisy given the proportion of productive edits to... not so much.

What'd probably work would be another tab on your profile (or a filter on your votes tab)

this could be coupled with a periodic notification. Like, "n posts you've closed/downvoted were edited in the last day"

That'd put an upper limit on the noise

@Shog9 - From the discussions on meta, that was the best suggestion I saw for it as well.

7:43 PM

@Shog9 would be fine with me

7:54 PM

^

2 hours later…

9:35 PM

@Derpy what other server? And why?

1 hour later…

Sonic the Inclusive Hedgehog

10:59 PM

@JoeFriend Can you please look into potentially fixing this bug?

@Shog9 You previously said you didn't care much about this bug, but it now has a much wider scope, covering all sites with responsive design, rather than just Meta as before. Is your position still the same?

« first day (2921 days earlier) ← previous day next day → last day (2083 days later) »

Transcript for

Aug '183

Tavern on the Meta

MY GOD, IT'S FULL OF STARS AND BACKLOG! General friendly chit-...

bug crickets discussion excuses support waffles

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile