Oolite Bulletins

Posted: **Sun Mar 11, 2007 1:47 pm**

Oolite test release 1.67.1 - Band-Aid Edition is now available from GURPO. This release primarily fixes bugs found in 1.67. The updater can be applied to 1.65 or 1.67.

Changes:

Advanced Navigation Array can now be purchased (for any ship).
Yaw control, now with workingness. (Keyboard only. Default keys: , and .; not available in strict mode.)
Hopefully fixed crashing bug in streaming sound.
Increased precision of certain geometric calculations, most noticable in laser accuracy.
Cache is now rebuilt if OXPs are added or removed. Cache is now saved when docking rather than when saving.
Fixed two similar sound glitches.
Fixed calculation of sound size; sounds larger than 1 MB will now be streamed from disk (was 16 MB for stereo and 8 MB for mono).
Introduced, and hopefully fixed, a bug that made mouse control randomly not work.
Made application about 11 MB smaller.

Posted: **Mon Mar 12, 2007 9:32 pm**

OK, being running this for about 90 minutes of gameplay now and I've found the following;

Advanced Navigation Array was available for purchase and seems to work fine.

Yaw control works fine.

No crashes so far (playing with custsounds in).

Not sure about the increased accuracy of laser fire but I would stress that I've never been very good in combat so I might not be the best judge of this.

The procedural textures look very nice but they can look, well, not round, at a distance. Actually, angular is probably the word I was looking for. It also causes a significant pause every time you enter a system but I'm assuming that is the known performance issue.

NPC ship behaviour;

I believe that I'm seeing far more ships with a clean rating opening fire on me, when I destroy them I, presumably obviously, don't receive any bounty and on the last occasion this happened, when police were within scanner range, I became an Offender and was attacked by them.

I also am seeing more ships firing Hardened missiles at me. I've also been attacked a couple of times by ships who fire a salvo of missiles, so far the missiles haven't been Hardened, thankfully.

That's it for the moment, if I think of anything else then I'll post it.

Posted: **Mon Mar 12, 2007 10:23 pm**

Have you enabled the streaming sound testing stuff like I asked over there? If so, you should be seeing a lot of console messages starting with ”[sound.streaming.releaseContext”. I’m particularly interested in “sound.streaming.releaseContext.deferring” ones, and whether they’re matched by “sound.streaming.releaseContext.deferred” messages.

Posted: **Mon Mar 12, 2007 10:46 pm**

Yes, I did. Don't get any deferring/deferred messages but do get a lot of messages similar to [sound.streaming.releaseContext.immediate]: Streaming sound <OOCAStreamingSound 0x5ca310>{"guiclick.ogg"} stopped with 0 pendingCount, releasing context.

Also get an occasional [strings.conversion.vector]: ***** ERROR cannot make vector from 'laser_color' because 'Found less than three float values.'

Posted: **Mon Mar 12, 2007 11:10 pm**

Well, the first one is what I get, and almost certainly not the context in which the crashes occured. Weird.

The latter is probably an OXP’s fault. It means what it says.

Posted: **Wed Mar 14, 2007 1:45 pm**

Code: Select all

[script.debug.syntax.action.badSelector]: ***** PlayerEntity DOES NOT RESPOND TO scriptAction: "pauseAI 4.0"

Perhaps when this works it would explain why my girlfriend complains about me taking so long to answer her.."Sorry dear, I was paused for a while there.."

On a more serious note, I'm getting seemingly random pauses where the game obvously works something out and then the action kicks back in. Seems to happen in combat, or possibly it's just not noticable at other times, and makes aiming rather difficult when it happens, due to overshooting your target.

Definitely getting salvos of missiles coming at me, largest salvo was 4 missiles, and am also getting more pirates using their escape capsule.

Edited for spelling.

Posted: **Wed Mar 14, 2007 1:57 pm**

Uncle Reno wrote:
On a more serious note, I'm getting seemingly random pauses where the game obvously works something out and then the action kicks back in. Seems to happen in combat, or possibly it's just not noticable at other times, and makes aiming rather difficult when it happens, due to overshooting your target.

I hope I’m wrong here, but this could be to do with the cache pruning mechanism. to verify, use:

Code: Select all

defaults write org.aegidian.oolite logging-enable -dict-add dataCache.prune -bool yes

…although that’ll increase any slowdown due to pruning, so what’s really interesting is whether you get these pauses when it isn’t pruning. (That said, I’ve come to realize that dynamic pruning isn’t necessary; pruning before writing the cache, which doesn’t happen in flight, should be sufficient.)

Mmmm… prunes.

The ”pauseAI” should be ”pauseAI:”.

Posted: **Fri Mar 16, 2007 9:03 am**

A "while I think of it" comment...

Yaw control, makes things a lot easier when lining up shots in the rear view as, unlike roll left/right, the control isn't reversed.

Also, IMO, the rate of yaw movement needs scaling for different ships, i.e. an Anaconda yaws at (seemingly) the same rate as an Eel. Don't know if this is possible?

Posted: **Fri Mar 16, 2007 10:46 am**

It appears as if you’ve got a point there. In the next release, max_flight_yaw will be settable in shipdata.plist, and will default to being the same as max_flight_pitch. The not-being-reversed thing is an illusion based on how our concepts of direction map to reality. :-)

Posted: **Fri Mar 16, 2007 1:41 pm**

So Oolite is reality then??

Another request/suggestion, can page up and down be activated for use on the shipyard screen? It's only a small thing but not having to use the cursor keys would be nice.

Posted: **Sun Mar 18, 2007 3:46 pm**

Ahruman wrote:
I hope I’m wrong here, but this could be to do with the cache pruning mechanism. to verify, use:
Code: Select all
defaults write org.aegidian.oolite logging-enable -dict-add dataCache.prune -bool yes
…although that’ll increase any slowdown due to pruning, so what’s really interesting is whether you get these pauses when it isn’t pruning.

Still getting the pauses, although they don't seem any worse.

Posted: **Mon Mar 26, 2007 1:39 am**

I’m both pleased and embarrassed to say that I seem to have fixed [some of?] the pauses/stuttering, although I haven’t tested extensively, what with it being tomorrow morning.

It seems to have been in the logging system. Specifically, the entire working-out-which-logcontrol.plist-key-applies-to-this-situation thing was being carried out every time something might potentially be logged, because although there was an efficient cache mechanism in place, the actual storage for the cache was never set up.

Having fixed this, the cache hit rate is well over 99%.

I’m not entirely sure why this extra work should affect the game as much as it apparently did, but I suspect it comes down to a few extremely verbose bits of logging happening when new ships spawn.

Posted: **Mon Mar 26, 2007 6:32 am**

Excellent. Does that mean we will see a 1.67.2 soon?

Posted: **Mon Mar 26, 2007 10:39 am**

Hopefully, there will be a version 1.68 this weekend. The primary feature will be rudimentary JavaScript support (probably as an alternative to script.plist only, and definitely without a fancy new scripting object model). This’ll be implemented less elegantly than my initial attempt, but will have the advantage of working. Hopefully this will be a simultaneousish Mac and Windows release; we seem to be lacking a Linux maintaner.

Other stuff that’s done:

Better error reporting for dodgy plists.
Better performance in cache and logging subsystems – marginal increase in sustained frame rate, significant improvement in smoothness.
Vertex and fragment shader support on Windows as well as Mac OS X; new “hull_heat_level” uniform; availability macros for uniforms.
Minor internal clean-up stuff (woo).
Slightly faster maths routines (expect speed boosts of up to 0.1% on PowerPC).
port_radius now works.
addCargo: with negative value has no effect, rather than semi-freezing.
Fixed version checking and mismatch reporting for requires.plist.
All log messages will have a message class more specific than [unspecified], although not necessarily much more specific.
Mechanism to prevent silliness in rare case of cache file being moved from little-endian to big-endian system now actually works.
Fix for “ don't know how to load entity…” messages on GNUstep systems.

Stuff that will probably be done:

That JavaScript thingy.

Stuff that might be done, but probably not:

OXP verification option. (Initially plists, later texture dimensions, references to files not in OXP or Oolite, documentation files inside .oxp directory…)
Fix for ships zooming past screen in demo mode.

Edit: moved entity loading error fix into “done” list.

Posted: **Sun Apr 01, 2007 8:54 am**

Don't know if this is the right place for comments on the latest test release (seeing as I'm a PC user and not a Mac user). However, the latest release seems far, far, far more stable than 1.65. Usually I'd play Oolite and have a crash within 45 mins. So far I've managed to play for 2/3 hours continuously without issue.

This is great work. Just one question - as I'm playing with the traditional Elite keys what would be the keys for the yaw movements?