Zugg Software :: View topic

Posted: Thu Sep 23, 2010 8:09 pm

I uploaded a quick fix with v3.29a today.

Should fix the issue of @db variables within regular expression triggers.
Should help more with slow walking issues
Fixes the latest issue with Reformat

Wizard Joined: 26 Mar 2008 Posts: 1547

Thanks Zugg!

Posted: Thu Sep 23, 2010 8:39 pm

Thanks Zugg. I can understand that the various configurations possible can be taxing to code for. I'm curious as to what was the issue with the slow walking and what you did to fix it. I've quite enjoyed reading your explanations in the past and they often help my understanding for new issues that pop up.

Posted: Thu Sep 23, 2010 9:07 pm

There was an internal flag called "OKToStep" that determines if the next line from the MUD causes the mapper to move to the next step in a slow walk. This variable was not getting reset properly. So sometimes it would be true when it shouldn't, which was causing #SLOW to immediately move, and was causing the next move after an #OK to move, etc.

And yeah, this mapper stuff is incredibly complicated. Doesn't help that a lot of the mapper configuration code (and slowwalking code) was ported from zMUD and some of it is 10+ years old. That code exists along side some of the newer code added in the v3 mapper rewrite. In the rewrite I tried to make the code more understandable but also didn't want to rewrite the configuration stuff yet (that was supposed to wait till Phase 2, but I've essentially done it now instead). So it's just a mess and making a "fix" in one place can easily cause side effects in other places.

However, in this case, turning on the GMCP stuff just made some existing bugs in the mapper more obvious. Some of these slow walking issues have actually been around for years. They existed with MXP and #TAG, but only got enough testing with the new GMCP stuff to become more obvious and reproduceable.

Posted: Fri Sep 24, 2010 12:42 am

I've found 3.29a to be rock solid in terms of walking around in SLOW mode. I haven't experienced a single glitch of any type in trying several very long walks.

On the downside it seems to have a serious problem with no onRoomEnter events being produced (see separate topic).