ThreadBoard ArchivesSite FeaturesActiveworlds SupportHistoric Archives |
Major bug report - confirmation needed (Community)
Major bug report - confirmation needed // CommunitybrantMar 13, 2001, 7:44pm
I've reported this bug to Roland several times, but this time the effects
have been more serious. Therefore, I thought I should post to the newsgroups to see if anyone else has encountered this strange bug that has shut down several of AWTeen's events. Below is a copy of the email I've sent (again) to Roland. --------------------------------- I reported a bug earlier (in late February) with lights that didn't seem to refresh. In case you don't remember, the bug occurred when a lot of people were gathered for a light show (about 35). The bots turned off the moving and rotating lights, but the lights didn't disappear, apparently because the objects weren't refershed correctly. Not only did this bug occur when there were a lot of people in the scene, but also when objects were being changed rapidly by the bots. However, the bug seems to affect all objects, not only lights. At the AW SURVIVOR Challenge II: Frozen Desert (AWTeen 1800s 1800e), between 60 and 70 people were gathered for a competition called wacky races. The event involves trying to move along poles (with shifting, flying, and local teleports disabled) to a finish line. The event had to be scrapped because of a bug in the AW browser. Apparently, when all the people were gathered there (some using 2.2), multiple duplicate fence objects had been made, overlapping one another (most likely by accident.) Anyway, those with 3.1 were able to move through the un-refreshed fences and go onto the course, while those with 2.2 saw the fences and weren't able to pass. My attempts to delete the fences with 3.1 failed, but I went to another computer that had 2.2 on it and it worked. Furthermore, later in the same event, someone was making a tournament ladder consisting of sign3.rwx's (which you can see at the above coordinates). I was watching the creation of the signs for a while, until I noticed that they were in disarray even though some others (who I have now determind to be using 2.2) could see them clearly. It turns out that several of the tic-tac-toe bots had been malfunctioning, issuing hundreds of AW_OBJECT_CHANGE commands per second to the signs. After I corrected the bot problem, however, the others who were present (again, about 50 people) were unable to see the signs, and I had to cancel this event as well. Before I canceled the event, however, I restarted the AW browser, and I could see clearly; unfortunately, I didn't connect the others' not being able to see the signs with a browser bug at the time. Most of the 30 or so 3.1 users who were present can vouch that these three bugs occurred. I would appreciate your looking into this major bug, as future events cannot be planned when these malfunctions will interfere. Thanks. If you require more information, please reply. My 3.1 system: 1.5 GHz Pentium IV Dell Dimension 3100 series Geforce 2 (64MB video RAM) - latest drivers 256MB RAM 56.1 GB free hard drive space (/60GB) Windows Millennium Edition My 2.2 system: 400 MHz Dell Latitude LS (laptop) - Pentium III No video card 128 MB RAM 4.1GB/10GB space free Windows 98 (second edition) joemanMar 13, 2001, 7:58pm
agent1Mar 13, 2001, 8:11pm
Object refresh rate simply dictates when the browser check the OP for new versions of the objects it has already downloaded. It has nothing to do with this problem.
-Agent1 [View Quote] syntaxMar 13, 2001, 9:40pm
xelagMar 14, 2001, 12:32am
I f these are xelagot bots, could you be more specific? How did you
ascertain that they were doing this? Xelagots do *NOT* issue hundreds of AW_OBJECT_CHANGE per second, as far as my knowledge of xelagots go (and I think I know quite a bit about them). That would be ineffective, and probably disconnect the bot inmediately, moreover, the scripting engine would not allow this. This seems to have nothing to do with the bug your are talking about, but I'm curious to know how you came to this conclusion... XelaG > It turns out that several of the > tic-tac-toe bots had been malfunctioning, issuing hundreds of > AW_OBJECT_CHANGE commands per second to the signs. brantMar 14, 2001, 1:20am
Oops, I overestimated. There were six tic-tac-toe bots running in the area,
which all went haywire (I don't know whether that was intentional or a bug) when people began clicking on the "Click for new game" signs. The script, apparently, doesn't prevent others from clicking the reset signs after someone clicks them the first time, and the bot doesn't switch back to "Click to register" mode until after the bot resets however many times the sign is clicked. For example, if I clicked the "Click for new game" sign six times in succession, the board would reset six times before displaying the "click to register" message. The bots, under observation, seem to take about one to two seconds to reset the signs each time, meaning that there could be anywhere between (9 (signs) * 6(boards) = 54 change commands per second) to half that (if it were once every two seconds = 27 change commands per second). In addition, several chess games were being held nearby, issuing AW_OBJECT_ADD and AW_OBJECT_DELETE commands, adding to the confusion. I'm almost certain that the bug has to do with bots in the area changing objects rapidly (not sure about the number of people, because I observed it when there were only two people present). So far, every racing event I've tried to hold with 3.1 has gone badly wrong, as they all involve bots that issue AW_OBJECT_CHANGE commands that operate gates, switches, etc (which is why I've had to scrap events I had planned for April until the bug is resolved). Not that this has anything to do with the AW malfunction, but I'll have to try to find some time next month to fix up that tic-tac-toe script to repair that bug. -Brant kahMar 14, 2001, 2:16pm
uh...you DID use unique names for all the T.T.T. sign? If not, you'll kill
most bots! And that racing bot, was it a self-written one (I've heard that you make VB bots...)? VB will get overheated by that many changes! Try to slow down the thing a little, don't have ANY other apps using your connection other than the AW browser and the racing-bots, if it's own-made use the fastest there is, C/C++ and don't use a Xelagot X1 bot, use an AV99Bot instead. Anyway, there are also limits to what the Uni and World servers can do. Well, you might want to try using these tips before scrapping any more events :-)) KAH PS. TEST the event concept before planning it! brant <awteen at shoemakervillage.org> skrev i meldingsnyheter:3aaee399$1 at server1.Activeworlds.com... > Oops, I overestimated. There were six tic-tac-toe bots running in the area, > which all went haywire (I don't know whether that was intentional or a bug) > when people began clicking on the "Click for new game" signs. The script, > apparently, doesn't prevent others from clicking the reset signs after > someone clicks them the first time, and the bot doesn't switch back to > "Click to register" mode until after the bot resets however many times the > sign is clicked. For example, if I clicked the "Click for new game" sign > six times in succession, the board would reset six times before displaying > the "click to register" message. The bots, under observation, seem to take > about one to two seconds to reset the signs each time, meaning that there > could be anywhere between (9 (signs) * 6(boards) = 54 change commands per > second) to half that (if it were once every two seconds = 27 change commands > per second). In addition, several chess games were being held nearby, > issuing AW_OBJECT_ADD and AW_OBJECT_DELETE commands, adding to the > confusion. > > I'm almost certain that the bug has to do with bots in the area changing > objects rapidly (not sure about the number of people, because I observed it > when there were only two people present). So far, every racing event I've > tried to hold with 3.1 has gone badly wrong, as they all involve bots that > issue AW_OBJECT_CHANGE commands that operate gates, switches, etc (which is > why I've had to scrap events I had planned for April until the bug is > resolved). > > Not that this has anything to do with the AW malfunction, but I'll have to > try to find some time next month to fix up that tic-tac-toe script to repair > that bug. > > -Brant > > brantMar 14, 2001, 6:08pm
Thanks for the info. However, I've never been able to figure out how to
display a simple window in C++, even though I'm experienced with DOS C++ programming, so the AW SDK wouldn't be of much help, unfortunately. The main reason I'm hesitant to try again is that this is the third event that I've had to cancel because of this bug, and I received a lot of greif over the prizes for the events that didn't go as planned (people can get pretty antagonistic when they're trying to win a free world - lol). Most of all, however, I don't want Lucrezia or other AW staff receiving telegrams of any more botched events that make the world's staff look incompetent. Maybe we'll just hold a trivia competition or something that doesn't involve objects until the bug is resolved. You sound like you would be interested in helping plan events - I'm always very busy and we could use some help anytime. |