@CodyGray @JeffSchaller For the "buy within 400 characters" watched regex which you were interested in, has, since it was added, 8375 Total / 3337 TP (39.84% TP) / 4885 FP / 136 NAA. The TP rate could certainly be improved. For this, like many watches, the important thing will be to make sure we're still detecting the TP where this is the only detection.
I could tell just from the regex that you put a lot of thought into it; I'm simply biased because all I can remember are the "fp" from it. But that's not something to optimize :)
Ahhh... That would certainly do it. :) There's no rush on my account. I was merely pinging you, as assumed it was functional and you'd want to know that it wasn't. :;
@Ollie That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body Append -force to the command if you really want to add the pattern you entered.
@Makyen That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body Append -force to the command if you really want to add the pattern you entered.
@forest Yes, that would be how you would report a spam post to SmokeDetector. It is helpful, but not required, to check first to see if it was previously reported/detected. If it was already detected, then SD will notify you of that when you attempt to report it.
If I want to add "www.onlinedocumentos.com" and its variations to the blacklist, would I just do !!/watch onlinedocumentos.com? And documentoregistrato.com as well.
@forest There are separate blacklists and watchlists, which are described in the documentation. But, yes, to add a domain to the watchlist, it's quite straightforward. The watch- and black-list items are regexes, so you'll need to escape the period.
@forest That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body Append -force to the command if you really want to add the pattern you entered.
@forest You can also use the "bisect" command to see if some text would be caught by an item on the watch- or blacklist. Note that when bisecting, you need the original text, not the regex, so don't escape the periods.
The test commands are preferred, due to bisect being compute intensive, if you just need to know if it's already caught. The bisect command is primarily for when you need to know exactly which regex in the watch or blacklists actually detected the text. If you are expecting to add something if it's not already on the list, then just go ahead and try adding it, as adding it will effectively run test and report all but a few detection types (e.g. ones which don't overlap text detections).
@Ollie As I noted elsewhere, the chick with part of his shell still stuck to his head, covering his eyes, was meant to be both a celebration of Easter and also of the unique perspective of the Stack Exchange design team. However, people were not recognizing me, so Ryan helpfully added the glasses (and improved the contrast around the shell, to make it look like the chick didn't simply have his head cropped off).
Now that Orthodox Jesus has also risen, it's probably time to revert back to the normal profile picture.
I am also considering whether we should break up the user blacklist commands into a separate top-level (H1) section, as they're almost entirely unrelated to the other watch/blacklists.
A program that buggy probably also has some serious security issues. I wonder if it's possible to cause arbitrary code execution from a post that it scans...
@forest I would be rather surprised if you could cause arbitrary code execution in a python program, but if you can, then please have a look at charcoal-se.org/security
Memory leaks, especially slow ones, can be very difficult to debug in production applications.
@forest I'd encourage you to choose your phrasing more carefully, then. "A program that buggy probably also has some serious security issues." really sounds like you're talking about SD, not Python.
I agree with it being easier to write buggy code by mistake in interpreted languages. But I'm not sure how an interpreted language makes memory leaks any easier to write.
@forest Why? Python's memory-safe. What makes an interpreted language inherently harder to hunt leaks in? Perhaps the profiling tools are worse, but that's not inherent to interpreted languages. If anything, being interpreted means that it has more of the program state in the interpreter in a potentially useful manner that could be dumped.
@RyanM Legitimate companies may hire a firm that appears legitimate but which actually spams, or which does so via a third party. Also, spammers may be spamming for companies they have a stake in.
@forest I'm no fan of Python either, as those who subscribe to my newsletter already know quite well. But I am a fan of the fantastic work that this particular Python application does, and, more broadly, of all the hard work that volunteers have put into this project.
It's obviously not perfect. There's plenty of room for improvement. And it almost certainly could be made better with a fresh rewrite, using the lessons learned.
Is the codebase particularly large? I would assume that something that just scrapes posts using an API, subjects them to regex, then does some ns lookups wouldn't be too complicated, but I don't know if that's all it does.
@SmokeDetector Looks like a spam seed: The question was asked and answered by the same user, with very little time in between, and the answer is link-only.
@Ollie I do not think so. These are legitimate projects, and they're not affiliated with each other. This is just a retrocomputing enthusiast attempting to build a "big list".
@CodyGray That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer Append -force to the command if you really want to add the pattern you entered.
