1. It required Python (and not just any Python, either- it had to be 2.3, which is older than current) to be installed before I could fire up the client.
2. It crashes on not finding anything in gamedata/particles, but there isn't any documentation saying what, if anything, needs to be in that file to make it go. After going through the SVN notes, I found AF's cryptic notes about making the following entries:
Code: Select all
[nanospray]
{
allow_team_colours=1;
}
3. The reflective water is gorgeous and much faster than the dynamic stuff, but it doesn't seem to handle explosiongenerator groundflashes correctly yet.
4. There's a a very weird aiming bug going on that's affecting certain units. I strongly suspect it's the ones with custom aiming arcs... where I wrote custom BOS code so that they'd aim properly in the first place- a hack for the last version of Spring. I'll test this assumption and report back. I assume this has to do with finally fixing the weapon-arc problems with things like the Strider, so that they steer towards their closest enemy, instead of charging right past them. However, without an AI build to test with, this is pure speculation.
5. I'd really like to see a NTAI build for this, so that I can do more practical load-testing than just building 500 Wolves and telling them to converge and attack a single point (which is hilarious btw).
6. I see we're now using the Flare graphic as the default graphic for flamethrowers, but the SVN, if I recall correctly, implies that this can be customized somewhere. I should probably read through those commits and find out how I'm going to test this nifty new particle code... and determine whether the big bugs with hit-detection, which seems to affect most of the weapons using FX (most especially beamweapons and beamlasers) have been resolved... otherwise this Great New Thing will actually be a big waste of time. Yes, I know... hitscan stuff is boring as all get out. The thing is, the different Weapon Types aren't all using the same code. The code for models works great, the code for things with FX does not. Personally, I think that the best way to resolve this would be:
A. All Beamweapons, Beamlasers, etc., on each firing event, fire TWO THINGS: a particle/fx object (our graphical effect) and a S3O point-object (included in a new revision of Base as, say, "point.s3o", just in case some modder really needs to alter it) with a Radius of 1.0. Both objects, needless to say, have to travel the exact same path, or this is all pretty pointless.
B. Now have all of these Weapon Types use the .S3O-models-as-weapons code to determine hits, using that point object, and cut out the now-redundant code for the special cases. Voila... now we'd have fancy FX and proper hit-detection, which would make us modders happy... and it would be one fateful step towards finally arriving at a unified Weapon Types model

7. We really should have the ability to draw ground patches beneath moving units. Then we could finally have fake "blob" shadows, or even fake-perspective raytraced ones, that would be faster than drawing raytraced ones. I'm not even saying I prefer this option to the current shadow code, which is quite pretty. I'd just like this to be an option, so that I could maybe offer an "NanoBlobs for the Computer-Impaired", or whatever, with fake shadows.
8. I see that the bug Caydr reported (and that I have supported, with screenshots) with the Origins of S3O models getting displaced during rotational motions has not been fixed. This is a serious problem on three levels:
A. Flying objects, as they turn sharply or spin, are getting the hitsphere displaced by as much as the original radius. After testing, I have determined that this happens when the hitsphere is not at precisely 0,0,0...
B. It is apparant that the real positions of the model's XYZ is not the position of the model's actual geometry! This makes very little sense... among other things, it implies very strongly that S3O objects are being, in fact, tracked at least twice. This is almost certainly because of the un-abstracted 3DO code and the areas where S3O was shoehorned in, so I guess we'll just have to live with it until a mighty coder can clean this mess up.
...that's it for now. I'm going to test the Friendly-Fire stuff and see if I can resolve the aiming bugs I'm observing with limited-arc units (again, this is probably caused by my hack-around for the last version, so I'm withholding conclusions until I'm done).
If the Spring developers would like, I will release an interim version of NanoBlobs using all of the new features as soon as I have it ready, for testing purposes.