On Tue, 14 Oct 2025 04:47:10 -0000 (UTC), Duncan wrote: > David Chmelik posted on Sat, 11 Oct 2025 02:29:04 -0000 (UTC) as > excerpted: >> Nevertheless, I tried GigaNews. They've been very helpful, though when >> I started using their server, a large number of posts' dates are wrong >> at beginning and end of some/most newsgroups. The beginning has >> question marks, and some before Usenet, and a few--not enough--after it >> started, and most since only 2003... then it has some from '2026' to >> almost the year 10000. What's most concerning is 1979 to 2003 gap. > > That's... interesting. Fair warning: This goes rather train-of-thought > as I try to come up with plausibilities... =:^) > > First thing I'd try is a look at the full headers, and if you don't see > it there, the raw message, either dug directly out of cache (should be > cached by Message-ID as file-name, .msg extension IIRC), or > saved-as-text and then opened in a text editor. Tho I'm not absolutely > sure that pan does save-as-text as a direct cache copy, vs. saving it > from the possibly processed state in RAM, so I'd not rely too much on > save-as-text particularly for any "funny" behavior like this that you're > trying to debug.
A rec.games.frp.dnd post after the year 8000 says in header it was actually posted in 2005... > (Written as I'm about to send, after thinking about all the > possibilities in this post... Now I'm curious!) I'd be interested in > seeing one of these raw messages, actually. Maybe you could post it as > a file attachment? Sure: how would I get an example raw post or few and should I just email you it? > If the date headers in the raw message are still clearly insane, then > you gotta decide whether the problem is likely to be a troll doing it > deliberately (or maybe just someone with a horribly screwed up posting > client!), or if it's likely to be a giganews issue, or possibly a pan > issue. It was Forte Free Agent in the year 2005... I still like that one (enjoyed it in 1990s) but by then was outdated and maybe has bugs? > You could try another client. IIRC even lynx (the text-based browser) > has news/nntp handling, and could possibly be used to fetch the message > for comparison. Or try some other more conventional news client. I will also be retrying Thunder-/Better-bird to see collapsible hierarchies to find faster... and I have so many in PAN now don't know how much longer I can keep up looking through those daily without collapsibility... but I don't want to lose the protection of not emailing some random newsgroup a personal email, which happened once, albeit only my new PC order to local PC sales/build/repair shop, so no big deal (not a message I sent a girl I like). > If a second client comes up with similar screwy dates for the same > message-id, then it's almost certainly stored that way on giganews. > Maybe first check their FAQ (if any) to see if it mentions screwy dates. > If not, at that point... it's giganews, maybe actually test their > support claims to see if it's worth the money. Presumably they'll need > the message-id at least, and the author, newsgroup, xref if possible, > and claimed date, may help. Or just send them the raw message > (including all that stuff and more) as you dug it up and see what their > support says they get by comparison at their end. Since it's correct in the header my suspicion is GigaNews messed up archive. > A couple things I do seem to recall, however. I believe the message > format may have been different say back in the 80s, before full > standardization. It's possible something screwed up that conversion or > whatever. Also, there are of course even still gateways (like the gmane > list2news gateway) and private servers (like the microsoft servers and > groups that at least back in the 90s weren't officially propagated, but > were mostly on the public servers too... in perhaps not complete form > but there). And I forget what it was called but there was another set > of "groups" that were actually converted/gatewayed as well, from IIRC a > bulletin-board format... Given that completeness is a giganews selling > point, it's quite possible they have perhaps not entirely accurately > converted some of these old formats from archives, and maybe that's got > something to do with the wonky dates. yes: likely > Another possibility is epoc dates, Unix (Jan 1, 1970, is day zero, and > "negative" numbers would get pretty huge if interpreted as unsigned... > also see y2k and 32-bit-y2038 messes) vs. something else. That > shouldn't appear in /nntp/ date headers, but if they're using a > date-stamped received date that could be in unix time... or they could > be converting from some non-standard or pre-nntp-standard format... > and/or pulling from unreliable archives... Maybe some messages don't > have their date header at all and the server /is/ using a received > date... possibly as pulled out of a wonky archive or conversion... Epoch? UNIX era, but Usenet started 1979 or 1980...? > Of course the nntp/internet-message RFCs standardize a date header, > including order, but localized date formats vary across the world and if > some clients screwed it up or the message is converted/gatewayed/from- > wonky-archive... Yes. I wish I could change PAN to display EU/UK or ISO dates; my family is European-/British-American (lived on both sides of Atlantic multiple times) and as a kid I quickly realised American date format (except ISO/ NASA) makes no sense rather than is confusing. > In terms of missing posts, there are people who set x-no-archive or > similar headers, and IIRC there's also an old header that defined > propagation limits (with "world" or some such the default and widest > propagation), but it could easily be two decades since I looked at that > stuff, so the details are long out of memory. But any of those would be > posts here or there, not gaps of multiple years without any posts at > all! I didn't see extra headers. _______________________________________________ Pan-users mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/pan-users
