dmap after FixGlobalTjunctions

[grayman]

What is dmap doing after the line "FixGlobalTjunctions" appears?

 

I have a 1-room 95-brush map that dmap hangs on at this point (it's working on entity 0). I've got it down to where if I delete a brush in the map, it completes. Add the brush back, it hangs. Pick another brush, any brush, delete it and I'm ok. Put it back and dmap hangs.

 

I know it's not a brush limit problem.

 

I'm curious as to what dmap is working on at this stage.

 

Thanks.

[Fidcal]

Don't know. It certainly sounds like some complexity of arrangement that dmap can't quite handle and that last brush breaks the camel's back. As you say, it's not a max brush number but dmap has just gone into a loop trying to process the arrangement you have. Normally I might suggest put a big block of caulk over half the room to test where the error is but it sounds like that would work anywhere.

 

Have you some particular area of special complexity like an open wooden stair or framework of some kind? If so you might try converting all that structure to a func_static to see if it helps dmap. I would delete all the mapname files like proc. cm aas etc before dmapping so they are created anew for sure.

 

I'll be interested to hear what others suggest on this one.

[grayman]

Don't know. It certainly sounds like some complexity of arrangement that dmap can't quite handle and that last brush breaks the camel's back. As you say, it's not a max brush number but dmap has just gone into a loop trying to process the arrangement you have. Normally I might suggest put a big block of caulk over half the room to test where the error is but it sounds like that would work anywhere.

 

Have you some particular area of special complexity like an open wooden stair or framework of some kind? If so you might try converting all that structure to a func_static to see if it helps dmap. I would delete all the mapname files like proc. cm aas etc before dmapping so they are created anew for sure.

 

I'll be interested to hear what others suggest on this one.

 

I backed up to a known good map version and came forward again, ending up with the same configuration. dmap likes it now. The only complexity in the added brushes is the prefab arch_stone_large01. But in debugging the problem, I had started with the bad map, deleted everything I added, except for this prefab, then iteratively deleted prefab brushes and dmapping again. When I had a single brush left from the prefab, dmap finally liked it. But if I added back a brush, dmap failed. If I deleted a brush anywhere else in the map, dmap was happy again.

 

A strange problem.

 

I was just curious about what loop dmap was getting stuck in.

[Mortem Desino]

I've said this in another thread, but the D3 process sometimes appears to 'hang' when it dmaps. Don't be alarmed, it's still working. It happens to me just about every time now with my more complex maps, in windowed mode, with DR running in the background.

 

I don't believe it's a loop more than it is just the renderer and soundscape losing CPU time and halting momentarily. Compiling is a pretty heavy process you know...

[Fidcal]

Yes definitely give it 5 or 10 or 20 minutes to dmap and pathfind as the map gets bigger and it may appear to hang during dmap but it is just processing without output. When you add many AI it also returns to the menus at some point and seems to hang but it is processing the pathfinding. I'm just so used to it that it didn't occur to me that it might be taken as a crash. Of course, sometimes it really is a crash.

 

It all depends on map size and you get to know how long your current map is taking. With my current map I enter dmap then I press all the keys blind needed to map it: console key, arrow up key (to recall the dmap line) home key, delete key (to delete the d of dmap) then Enter. I can hit those 5 keys rapidly and automatically now. Then I go away and make a coffee, maybe find something around the house to busy myself with; come back in 15 to 20 minutes.

[grayman]

Yes definitely give it 5 or 10 or 20 minutes to dmap and pathfind as the map gets bigger and it may appear to hang during dmap but it is just processing without output. When you add many AI it also returns to the menus at some point and seems to hang but it is processing the pathfinding. I'm just so used to it that it didn't occur to me that it might be taken as a crash. Of course, sometimes it really is a crash.

 

It all depends on map size and you get to know how long your current map is taking. With my current map I enter dmap then I press all the keys blind needed to map it: console key, arrow up key (to recall the dmap line) home key, delete key (to delete the d of dmap) then Enter. I can hit those 5 keys rapidly and automatically now. Then I go away and make a coffee, maybe find something around the house to busy myself with; come back in 15 to 20 minutes.

 

I understand it can take long to dmap a map. My 2200+ brush map takes about 4 mins.

 

However, when my 90-brush map sits in dmap for 20 mins, and it normally takes 5 seconds, it has to be something other than loss of CPU cycles.

 

It has to have something to do with the complexity. I ran into "one more brush" hanging dmap on 2 separate occasions today. When I finally changed a complex catwalk into a func_static, the problem appeared to go away for good.

[Mortem Desino]

Oh yes... When the compiler turns brushes into loads, and you've got complex intersecting worldspawn, it will no doubt choke. I can't say that I've encountered a problem quite like yours.

[Fidcal]

yes someone else had this problem with an open wooden stair. I changed it to func-static then it was OK. Remember of course you then need monster clip on your catwalk if AI can get up there but not if they can't.