« first day (2899 days earlier)      last day (30 days later) » 

1:53 AM
@Natty tp
@Natty tp It is better to use .profile for this, but the post doesn't say how, it's really just link only answer, and also it's part of a wave of posts by that user that promote that site and should very likely be considered spam.
@Natty tp
@Natty tp
@Natty tp
@Natty tp Appears to be a success report.
@Natty tp
@SmokeDetector Should this be edited and undeleted?
@EliahKagan hmm that question must be dupe of something...
@EliahKagan that does seem likely to help someone
2:23 AM
[ SmokeDetector | MS ] Manually reported answer (batch report: post 1 out of 5) (94): Installing Current nodejs version by Anna on askubuntu.com
[ SmokeDetector | MS ] Manually reported answer (batch report: post 2 out of 5) (94): Command 'gradle' not found. How can I run the gradle command from any directory? by Anna on askubuntu.com
[ SmokeDetector | MS ] Manually reported answer (batch report: post 3 out of 5) (94): Maven can't read my JAVA_HOME correctly by Anna on askubuntu.com
[ SmokeDetector | MS ] Manually reported answer (batch report: post 4 out of 5) (94): apt-get install "maven" or "maven3" by Anna on askubuntu.com
[ SmokeDetector | MS ] Manually reported answer (batch report: post 5 out of 5) (94): Setting java path in maven by Anna on askubuntu.com
2:49 AM
@Zanna I've edited the post and raised a custom mod flag requesting its undeletion.
No problem. :)
Last night I was thinking, what can be done to increase the number of answers (or answered questions) on Ask Ubuntu?... I didn't come up with anything. But this morning I am trying to write an answer. That seems like a reasonable approach XD
Do we have a drop in answers?
(unrelated) Do we already have something on meta that addresses all the aspects of this newly posted meta question about review comments?
I don't think so... I was trying to think if there is some convenient place with the text of those comments
2:54 AM
There's actually two related questions now. Here are oneboxes of both, to avoid confusion:
Q: Post Reviewing Useful Comments

Sasuke UchihaI have seen the same set of comments being used by many reviewers. Can an experienced reviewer please provide a list of those comments? Here's one: While this link may answer the question, it is better to include the essential parts of the answer here and provide the link for reference. Link-...

Q: Link to Comments and Review Queue

Sasuke UchihaHow to get links to comments and to the review of a post to be added as "By Review"?

the canned comments aren't site-specific are they?
No, they're not.
Those two questions seem like they're asking the same thing.
yes... at least, the full answer to one will fully answer the other
@EliahKagan I had not observed one, I was just thinking that not enough answers is always a thing here. But I just checked the answered rate, which was, I think, 56% or 57% in 2016, and is now 65%, which seems an astronomical improvement.
is that question answerable? seems quite unclear to me
I think the question should be considered unclear.
@Zanna Regarding the answer, I'm not sure if that's really an answer, as written. If it is, it may benefit from an edit.
I voted to close the question
dupe confirmed by the OP's comment
3:23 AM
abandoned questions: 1 2 3 4 5 6
2 hours later…
5:49 AM
OT Android
[ SmokeDetector | MS ] Bad keyword in answer, blacklisted website in answer, pattern-matching website in answer, potentially bad ip for hostname in answer, potentially bad keyword in answer, +1 more (261): What is so dangerous about --trim-sector-ranges? by johna tterson on askubuntu.com
1 hour later…
6:59 AM
[ SmokeDetector | MS ] Bad ip for hostname in body, bad keyword in body, bad keyword in title, blacklisted website in body, link at end of body, +5 more (693): New Life Keto : A Healthy Way A New Quick Weight Loss by jameishwartions on askubuntu.com
1 hour later…
2 hours later…
11:03 AM
@EliahKagan Does it make sense to close this one as a dupe of this old question?
12:44 PM
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, body starts with title and ends in url, link at end of body (392): Your deck may seem by Solo Org on askubuntu.com
1:23 PM
[ Natty | Sentinel ] Link to Post Contains Blacklisted Word - my question; Low Length; No Code Block; IntelliBL - 0.2; 3.5;
@Natty tp
@Natty Is this intended as an answer or as a comment on the other answer?
Never mind, it was posted earlier than the other answer.
@Natty fp
@Natty tp The version Natty saw is definitely not an answer.
@Natty fp It's giving was is intended as a solution.
@Natty tp
@pomsky I think probably not, since the comment it quotes is an LQP review comment rather than a pro-forma userscript comment.
[ Natty | Sentinel ] Link to Post Low Length; No Code Block; Low Rep; 1.0;
1:33 PM
NAA, doesn't address the question that was actually asked.
(Including information about other OSes is fine, but none of the information in that answer, including the Ubuntu part, is answering the particular ROT-13 question that's asked there.)
@Natty tp
1:57 PM
2:29 PM
@Zanna It has been undeleted.
@Zanna Is this because we are doing better, or just because that metric has become a target and we're doing things that improve the metric itself without necessarily providing answers or other useful improvements?
Obviously it doesn't have to be, and probably isn't, one or the other. But that's the gist of what I'm wondering. In particular, if we're doing better, we may be able to do even better by doing what we've been doing but more so. Otherwise, that might not be the best approach.
@Zanna It's possible you're giving yourself too little credit, as I believe you suggested a tag wiki project in response to my mentioning that I'd been seeing very few tag wiki edits in the suggested edits review queue. I do very much like the idea of a concerted effort to improve tag wikis.
(Also, if I'm getting the history wrong and did actually come up with the idea, then I apologize. The chat history about this is somewhat confusing and interspersed with other discussions. But I think the idea of having an actual initiative to address this is yours. My contribution was, IIRC, to lament that users with insufficient rep to eit tag wikis directly weren't suggesting more tag wiki edits.)
I think it would be good to find posts that use code fences incorrectly, such that the first line that is intended to be displayed is instead hidden by being unintentionally written as a language for syntax highlighting. The sort of thing I'm talking about is, for example, how
```ek@Kip:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 18.04.4 LTS
Release:        18.04
Codename:       bionic
gets rendered as
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 18.04.4 LTS
Release:        18.04
Codename:       bionic
without the first line
ek@Kip:~$ lsb_release -a
@EliahKagan hey i'm very new to debiam linux please answer my question if you're free
@EliahKagan and should therefore be fixed to:
ek@Kip:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 18.04.4 LTS
Release:        18.04
Codename:       bionic
Or, better:
ek@Kip:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 18.04.4 LTS
Release:        18.04
Codename:       bionic
@RafaelNadal Hi. You may want to ask in a room where that's on-topic, or, better, to post the question with full details on a site where it is on-topic. This room ("Raiders of the Lost Downboat") is for curation / community moderation of stuff on Ask Ubuntu, including improving posts, closing and reopening questions, deleting posts that are submitted as answers but don't attempt to answer a question, and so forth.
For general discussion of Ubuntu, there's the Ask Ubuntu General Room. It's possible to ask for help with Ubuntu there too, though usually people will ask you to post a question on the site, which is frequented by far more people than any chat room. Once a question exists, it can be iteratively improved and updated when new information is available, including progress made with help from people in chat.
However, your question is about Debian rather than Ubuntu. Questions about Debian (except those that are also about Ubuntu, such as a question about setting up an Ubuntu system to dual-boot with Debian) are off-topic on Ask Ubuntu, but on-topic on Unix & Linux. Your best bet may be to ask your question there. If you do, you should make sure to describe your problem as thoroughly as possible.
Unix & Linux does also have a general chat room, /dev/chat, where you may be able to get some help with your question, but the folks there tend to be even stricter about having a question posted on the site first. If you do post a question and if you want me to take a look at it, please feel free to ping me in chat, but be aware that it might be a while before I get to it, and also I very well might not know how to help.
2:59 PM
Thanks sir
3:09 PM
3:54 PM
@Natty tp
4:27 PM
@EliahKagan I was not aware of any particular attempt to "improve" that metric, so I was totally surprised when I saw the huge difference. Earlier while I was kneading dough or something, I was thinking I would reply to my own message saying "... but given the extremely low resolution of this data, we cannot conclude that it is an improvement in the quantity or proportion of useful stuff"
@EliahKagan you had some greeeaaat ideas about the APT wiki
@EliahKagan oh good idea... there must be lots of those already
@EliahKagan "and the downloaded uTorrent file is right before you" is a nice turn of phrase.
I agree. It's not my phrasing, though -- that was in the original.
haha yes I saw it
@Zanna I don't know how to query for that.
According to this post, SEDE doesn't support regex.
5:08 PM
The issue is that with LIKE, which does a form of pattern matching much weaker than regex (comparable to non-extended globbing, though with a different syntax), I don't think it's possible to have a subpattern match arbitrarily many characters characters from a (non-universal) character class.
Using a fictitious * operator with the same meaning as * in regex, I could write
Body LIKE '%[`~][`~][`~][^' + CHAR(10) + ']*[^a-z' + CHAR(10) + ']%'
@EliahKagan someone posted a query to find code fences
@EliahKagan I have run into that problem :(
@EliahKagan which I expect would have a manageable number of false positives. But LIKE patterns in T-SQL don't have anything like that that fictitious *.
@Zanna but that would obviously not be enough...
Is this the query for code fences? data.stackexchange.com/askubuntu/query/1023600/…
Searching SEDE, I get nothing for code fences and only that query for code fence.
no... but I can't find that query
but the one you found looks good
5:15 PM
Not for finding broken code fences, though.
That just searches for a newline followed by common amounts of indenting whitespace (including the no-whitespace case) followed by triple backticks. It conspicuously omits the case of triple tildes, but otherwise seems reasonable for that purpose.
Though if I were looking for code fences, I think I'd just look for triple backticks and triple tildes anywhere in a post. Triple backticks can also be used for inline code formatting, but in my experience posts this is only occasionally done in cases where it is actually reasonable, and any post with triple backticks that don't start a code fence is probably worth inspecting, since it's a heuristic indicator of other bad formatting.
SEDE seems to always forget my session. I have to login every time I restart my browser.
I have to log in even more often than that.
@jokerdino me too, and then I have problems getting to the login screen, or problems after I log in
I mean, I enter my credentials, but I'm not logged in
maybe it doesn't like me
5:19 PM
I'm strongly suspect it is possible to use SEDE to find broken code fences, at least for a smaller site like Ask Ubuntu, by using STRING_SPLIT with a newline as the separator and using LIKE on each resulting string. I don't know how to put that together into a query that returns one row per matched post, though, since I don't actually know SQL.
maybe we can do it together somehow
I don't think I know anything about writing SEDE queries since I just patch together bits of other people's code and fiddle with it until it seems to do what I want
but I do enjoy poking at it
5:34 PM
I found this query, probably not much useful -- data.stackexchange.com/askubuntu/query/488893?year=-1
that person would know how to write our query, if it is possible
@EliahKagan I have just discovered that I was totally wrong about the answered rate being lower in 2016. Back then, it was 65%
it then went up a bit
now, it has returned to 65%
5:50 PM
We can probably download the entire text and run regex locally
you mean, of all the posts caught by the query you found?
@Natty tp
@EliahKagan Body LIKE is probably not correct for this. I believe that looks at the cached HTML of the current revision, but we need the Markdown of the current revision.
Hence why this query (that I know is a draft) turns up posts in which triple backticks are actually displayed when viewed, rather than those that have code fences.
haha I was just saying that
I am looking for the queries I have that search the markdown
6:05 PM
@Natty tp
@Zanna PostHistory?
I'm going by how this query searched the Markdown.
So is the current idea to download information about all posts that appear to have code fences as a .csv file and then search that file locally?
shrug I haven't got an idea :D
@EliahKagan yes, you have to look in the post history table for the markdown
Anyway, I fixed one.
The cached HTML was fine but you can see the problem when editing it.
hahaha run the secret command
6:19 PM
The empty command in the sudo dhclient language.
What's that supposed to find?
I don't know.
I was just drinking water.
some chars after 3 backticks
6:29 PM
drinks water
30 degrees at midnight is too hot
Can I ping someone with an @ in comments if the user has edited the post previously (but not the latest edit)?
@pomsky Yes
@jokerdino Thanks :) also thanks for the "downloader" tag synonym
@pomsky thank you actually
6:40 PM
@jokerdino you're welcome then :D
@pomsky :)
@Zanna Did that take a very long time to return results when you ran it?
I think it was just slow momentarily.
For me.
I get about 12000 results
6:47 PM
but this is not useful
I am not sure I know how to correctly match a newline with LIKE.
@EliahKagan I have added the text to the output, and downloaded it, but, it seems I have thrown away the whole usefulness of the database in trying to do anything with it
I am running into a problem that I think might be that (some?) newlines are represented as CR LF instead of just LF.
what is happening?
6:53 PM
I expected this query, to return very few results, most of which had what I was looking for, but instead it returns hugely many, most of which didn't.
To be clear, that is not by itself useful, since most of the time people leave out a newline after the line that opens a code fence, the character that rarely appears in a correct name of a programming language is not the first character.
But I think it would be useful to figure out why that doesn't work the way I expect.
At first I thought it was returning results with trailing whitespace (which would be a false positive, though trailing whitespaces on the line that opens a code fence probably suggest that the formatting of a post is worth reviewing anyway). But that doesn't appear to be the case. It seems to turn up (at least) all posts that have code fences.
CR LF line endings appears to have been the problem.
But there are still a lot of false positives where it's not actually a code fence at all.
I get 1108 rows, which is about half as many as before, by requiring that it be on a new line (but I believe this also leaves out results when the problem line is the first line of the post):
WHERE ph.Text LIKE '%' + CHAR(10) + '[`~][`~][`~][^a-z' + CHAR(13) + CHAR(10) + ']%'
I'm going back to just backticks, since it looks like many of the remaining false positives may be things like
Nope, fewer than a hundred were. I get 1023 rows with:
WHERE ph.Text LIKE '%' + CHAR(10) + '```[^a-z' + CHAR(13) + CHAR(10) + ']%'
Well, many of these actually should be edited, though. In a lot of cases, people appear to have used triple backticks before and after text, all on a single line. Presumably they intended it to be formatted as a one-line code block, but it doesn't do that; it format it as a code span instead.
7:21 PM
unclear, and doesn't seem "suspicious"
@EliahKagan nice, will take a look tomorrow
two wrong/misplaced answers posted to the same question:
(related) dupe
@Zanna I've edited several. There are still over a thousand more. Also, I'm not sure edits are justified on all of them. I skipped one that was closed where it was readable, and one where there was no harm to the way it was displayed and no significant improvement to make.
I don't mean the kind jokerdino edited, where it was a code fence that sometimes rendered incorrectly but not in the cached HTML -- those should be edited -- but where it wasn't a code fence at all and the triple backticks were used in a way intentionally supported by SE.
@Natty fp This may benefit from editing, but it appears to have successfully diagnosed the problem. (Also, it's not hard to infer from it that one possible solution is to free up space, so even without editing I do think it qualifies as an answer.)
@pomsky Couldn't that be answered by explaining why it happens?
Is that a message that is sent to stdout by a postinst script, and thus shown to the user through the update notifier?
@Zanna My actual idea is to generate a long and horrible disjunction of LIKE expressions, one for each initial position of a probably-wrong character in a line that introduces a code fence, and paste them into the query.
The effectiveness of this approach is limited by how long SEDE will allow the query to run. It works for at least five LIKE expressions. I don't think that's sufficient, because the first questionable character could well be after that. Even with 10, it's timing out.
Even with 5, the query takes about a minute and a half to run.
8:05 PM
@EliahKagan OP made a strong claim: "I don't even have Dropbox (I don't want Dropbox) and I can't remember installing anything that requires dropbox integration." (which sounds very unlikely to be completely correct) and didn't respond to the commenter's suggestions and queries yet. Perhaps waiting for some more time to see if they reply would be reasonable too.
@pomsky You're right. I've gone ahead and voted to close it as unclear. We can reopen it if they give more information.
@EliahKagan I've managed to run it with 7 (which is actually an 8-way disjunction--it finds a character that doesn't likely appear in a language name at positions 0, 1, 2, 3, 4, 5, 6, or 7). I'm seeing if it works with more. This is a really hacky approach. 7 took almost two minutes.
It works with 8 (a 9-way disjunction) but times out with 9 (a 10-way disjunction). This might be good enough... but a better solution would be good.
Anyway, here's the code that generates the matching code: gist.github.com/EliahKagan/3d764c8099c7b652e1c2c01ee02dadaa
9:06 PM
I've edited some of them. In hindsight it would've been good to start with the most recently active posts rather than the least recently active posts! I'll try and do that whenever I pick it back up.
@Natty fp

« first day (2899 days earlier)      last day (30 days later) »