dengine 1.8.6 (latest stable) - problems with loading models
This question is mainly directed to DaniJ, but I would appreciate any answer from others.
Some time ago it was mentioned on NewDoom forums that deng 1.8.6 have got some huge problems with model loading engine, and that problems shows up in that each and every model that is set to be loaded (by specifying pk3 with -file parameter or by placing pk3 with it in Data/{game}/Auto folder or by placing it somewhere in the deng filesystem structure and providing appropriate references for it in ded file) might be skipped by dengine without any obvious reason. Even more, this failures might be rather sporadic (i.e. random) so say on the first run model gets loaded by engine but fail to load on second run and so on.
The question is: what are the details beneath the scene (on a code level)? Why does it happen? Of course it is possible for me to do some debugging myself, but it would take much less of my time it you will give me any insight to why does it happen.
Might this problems be related with a bunch of different pk3-s? Can they be avoided if there will be only one pk3 out there with all "user favorite" addons packed together? Can they be avoided If I would use no pk3 at all (i.e. if I will depack contents of each and every pack directly to deng basedir)?
Thanks in advance for your answer!
Some time ago it was mentioned on NewDoom forums that deng 1.8.6 have got some huge problems with model loading engine, and that problems shows up in that each and every model that is set to be loaded (by specifying pk3 with -file parameter or by placing pk3 with it in Data/{game}/Auto folder or by placing it somewhere in the deng filesystem structure and providing appropriate references for it in ded file) might be skipped by dengine without any obvious reason. Even more, this failures might be rather sporadic (i.e. random) so say on the first run model gets loaded by engine but fail to load on second run and so on.
The question is: what are the details beneath the scene (on a code level)? Why does it happen? Of course it is possible for me to do some debugging myself, but it would take much less of my time it you will give me any insight to why does it happen.
Might this problems be related with a bunch of different pk3-s? Can they be avoided if there will be only one pk3 out there with all "user favorite" addons packed together? Can they be avoided If I would use no pk3 at all (i.e. if I will depack contents of each and every pack directly to deng basedir)?
Thanks in advance for your answer!
Comments
If you open a pk3 file from the jDRP you can see, that all the ded's are directly in that pk3, not in "'pk3-file'\defs\jDoom\Auto". That's why the ded's are skipped (that's what I think).
If you unpack all pk3 files completely, put all the ded files into "defs\jDoom\Auto" (models of course as they are in "data\jDoom", so you can simply place that existing folder into your doomsday base directory, because the models are already in "'pk3-file'Data\jDoom") it should work correctly.
Provided you are using the correct version of jDRP for the Doomsday release you are using you shouldn't run into any problems.
There was a problem with loading models on Windows when using 1.8.6 in conjunction with Snowberry 1.0 but that problem has since been resolved by Snowberry using a response file to launch Doomsday.
In addition there was another issue in Doomsday 1.8.6 which prevented some models being loaded due to an algorithm for managing addons falsely identifying duplicates (name CRC collisions). This issue has also since been resolved.
Thanks for explanation, but the things you say are obvious - I had read all the discussions about in on NewDoom, and also Snowberry sources are very useful sources for invaluable information.
That's it! Thanks a lot! This is just the answer for my question I've been expecting to get. Now I know where to look at to patch 1.8.6 release for myself so it wouldn't miss some models due to a bug. Well, just as I thought, bugged one is the m_filehash.c ;-D.