On 2021/09/13 18:26, Simon Lees wrote:
Unfortunately I don't know if we can set up anything that complex on the list server,
Can't set up anything that complex....on a computer that can run can beat humans at running chess, among other tasks that have typically been only associated with human "intelligence"?
on the other hand adding a regexp to the list spam filter is pretty simple. Is it 100% the best solution probably not, is it the best we can do with the tooling we have probably.
You do know that you are allowed to run more than one tool in a chain? If email comes in for a list, it can be funneled to a socket where a tool can check the Regex for special processing instead of calling SA. SA can be called on other messages, but can be avoided for messages meeting any other criteria. =========================================== On 2021/09/14 01:08, Per Jessen wrote:
There are three conditions to be met for such a mail to be rejected:
if "subject matches regex" and "list is factory" and "sender is not dimstar" then reject.
So just changing 'T' to 'F' in Fumbleweed would allow it, ...or adding a space at the end, ...or using an alt for the listname (would have to investigate that, but it seemed like one of them worked), ...or putting dimstar's email address in the From line ...or...probably many others... And before anyone says any of those would not be ethical or a good thing to do -- we are talking about a response to something that is certainly, NOT a good thing to do and is also of questionable ethics, though most certainly, I think the actual reason at work, here, is laziness (including most of the above "workarounds", but they are in response to a particular "solution"). Certainly telling a random Jane that their response about release versions is 'spam' is unquestionably wrong, and may be worse than no solution at all. Using the [il]logic often used on me, if we allow or do 'X' for 1 person then we have to expect that ALL will do it. Right....you expect that everyone will respond to such a problem by trying 'N' different ways, as well as working to find out why, as well as pointing out what a horrible method that is, and how many other solutions could be used or made available? And no, you don't leave the current method in as a stop-gap until a better one is put in place, because if it isn't producing an annoyance or pain to whoever implemented the stop-gap, then even if emotional/personality factors prevent completion of a better solution, then environmental factors (other things needing to be done) will _seem_ more important. HOWEVER, if the irritant is left in place, then it's importance, relevant, to other issues won't necessarily allow for it to be ignored. What might be the easiest is that if the Regex matches, then check the body for some likely words, like 'bug', crash' blank-screen, "won't start", restore...etc, and send them a response with a link to the bugDB, with various, likely, fields filled out (proj: TW, ver: the date of the TW release (or one of the 4 I mentioned :-)), etc. Tell them to fill out other help fields, like reproducer etc. If it doesn't have any of those words in the body, send it to the list for triage 1st instead of bouncing it as spam (unless body contains references to Uganda, prize, etc). It doesn't have to be perfect, but it certainly would be be better than tossing it back as spam which had me looking at my what might have changed on my local setup and no clue that it might be something on opensnoozy's end.