Makyen

!!/standby Makyen

!!/location

@Cow You are getting better. It's a process. There's a significant learning curve. You're doing well.

@Cow np. Such things happen. I've definitely had similar things happen from time to time.

@Cow They appear identical to me: 32 Total / 32 TP (100% TP) / 0 FP / 0 NAA and 32 Total / 32 TP (100% TP) / 0 FP / 0 NAA

@Cow Your group of (?:auto[\W_]*+pay|autopay) is equivalent to just auto[\W_]*+pay, as that also matches "autopay".

@Fastnlight No [\W_]*+ between the "e" and "r"? Also, if they are going to go with that level of obfuscation, then you should probably also have [\W_]*+(?:d[\W_]*+o[\W_]*+t[\W_]*+)?c[\W_]*+o[\W_]*+m to account for the possible use of "dot" instead of ".".

!!/blacklist-number- 0879999148

!!/blacklist-number- 879999148

@Cow In SE search? Not that I'm aware of, but there's at least one possibility I'd try. However, without knowing there's a real match, it's difficult to verify. So, I created a few answers in MSE's Formatting Sandbox, which I used to verify the possibility I was going to try didn't work. As a result, using SE's search page, the only way I'm aware of is to manually, or write code to programmatically, look at each of the entries after doing the "paytm" search. SEDE might be able to help, if you want a limited search, as a full search of the body, at least on SO, will time out. If I was doing it, and I didn't just give up after looking at a few example results, I'd probably try an SE search first, then figure out which sites have hits, or at least the main portion of the hits. Then, I'd use the SE API endpoint /search/excerpts, which is the one that will search in both questions and answers, searching for "paytm", including the body in the results. Two searches per site would be needed, one for the text in titles, then another for it in bodies. I'd then take those results and search for "paytm=" (which could be done programmatically, or just in the browser from results on the SE API example page. Basically, it's a pain in the rear.

@Cow Why are you setting the "s" flag here? A) The "s" flag is already set for the watchlist and keyword blacklist. B) You're not doing anything that the "s" flag affects. So, I don't see the purpose in you explicitly setting the "s" flag here.

@gparyani Well, that should be interesting, and is not unexpected, given their plans.

@VLAZ Yeah, I assumed that what was provided by ChatGPT wouldn't work, but from the fact that you even asked ChatGPT, it wasn't clear to me that you wouldn't try to use it or some variation of it. However, I generally don't consider ChatGPT output to be worth the time to dig into in detail, so while the . and the formulation was concerning, from multiple points of view, I intentionally avoided spending time thinking about it (although glancing at it did result in me starting to do so).

@ipodtouch0218 We only see the mod flag that's active at the time we either load the flag dashboard or the question page. The ones that are self-removed we never see, unless the mod is, for some reason, looking at your list of flag, but doing so is quite rare. If you self-remove the flag while we're viewing the question page or flag dashboard, we won't know about it, but the next time the question page or mod flag dashboard is loaded the new flag will be shown.

As to detecting/not detecting arbitrary repeated text, that's certainly something that can be done with regex. Generally, it requires using a capture group for what might be repeated and then making sure what was captured is not repeated. Obviously, within a watch/blacklist entry it requires using a named capture group, but it's definitely doable. IIRC, there are a couple of entries which do use one.

@VLAZ Given that the only issue is repeating "cat", I'd probably use (?:(?:[CMX]AT|ATMA|CMAT)(?<!cat(?=\W+cat))\W+){4,}, which misses all the FP. It also misses 2 TP that are just "cat" repeated, but those are caught for other reasons. 14 Total / 14 TP (100% TP) / 0 FP / 0 NAA. I've made that change.

@VLAZ Don't use that as an entry, as the (*FAIL) assumes that the entry is the entirety of the regex. Thus in cases where the regex engine encounters the (*FAIL), none of the watch/blacklist entries after this one will be tested. In other words, the entire regex will be failed on that starting character and the regex engine will move to the next character. Hmmm... we should probably require a -force for using that directive, and, perhaps, a couple others.

The not reconnecting issue, which I assume you're referring to, is not the issue I'm looking at.

!!/location

!!/errorlogs 5

Still some problem with that

!!/errorlogs 10

@SmokeDetector No, it wasn't.

!!/dump_data

@Cow Welcome. :;

@JeffSchaller :)

@Cow A search in here for "long reply" produced:

@GeneralGrievance Yeah, this is a standard format for some shows someone I know watches. I think it's the history channel, but I'm not sure. It seems to basically double the runtime of the show. Before a commercial break they explain what they are going to tell/show you after the break. Then, after the commercial break, they basically repeat that preview/foreshadowing, then actually show/tell whatever it was they were going to show/tell. I find it very frustrating, and that's on top of the shows running at normal 1x speed (I normally watch shows at anywhere from 1.5x to 3.5x, depending on the content, with most in the 2x to 2.5x range). When that person has the TV on, which is almost all the time that I've seen them at home, I usually try to just ignore it.

@Cow You can use a long reply for those: 19 Total / 19 TP (100% TP) / 0 FP / 0 NAA

!!/standby Makyen

!!/pull

sigh

!!/watch- upstaff(?!\.com)

!!/watch- upstaff\.com

!!/watch- chemenggcalc(?!\.com)

!!/pull

sigh

@SmokeDetector Hmmm... That's not the commit that should have been pulled, as it hasn't passed CI, yet. So, we have a timing issue wrt. when MS sends a WebSocket message saying the CI testing is complete and SD pulling. I expect we need to have a check for the current HEAD having actually passed CI testing, as we do for the !!/pull command.

@SmokeDetector s/link/line/

@SmokeDetector Yeah, that's expected.

OK. It wasn't so critical as to need to happen right now. It would have been fine to leave it in for a while.

@Dragonrage np. Thanks.

@Dragonrage For performance reasons, please use non-capturing groups, (?:), rather than capturing groups, (), unless your regex requires using a capture group for functional reasons. If your regex requires a capture group, then you will need to use a named capture group, rather than an unnamed group, due to not knowing what other capture groups are being used in the full regular expression that is made up of all the entries on the blacklist or watchlist.

!!/blacklist-number- 127464426126

@JeffSchaller Sorry, I thought you were aware, yes, that's my assumption too. It's happened multiple times over the last couple/few days. However, it only started happening since @ThomasWard updated the underlying system, so it's possible that had some impact. But, I'm betting on Cloudflare.

!freeze

!freeze

!freeze

!freeze