Release Oolite

levi

Still fresh, damnit!
Joined
Oct 6, 2008
Messages
11,373
Location
Somewhere off the coast of the EU
There's rarely a smoking gun in the pndrun.out; it's usually moaning about sound buffer underruns at some point, then it's just either Segfaulted or more commonly plain Killed.

I did discover I was actually running a load of OXPs still. Not sure where they live in the appdata - I had the font oxp in my AddOns folder and I removed that, but in the manage extensions menu, it suggested I was running about a dozen. I've removed those now.

Just rebooted in case that was causing any weirdness. First on trying to run it without swap it fell over trying to load my old commander or create a new one. With swap, I was able to load my old commander, and launch out of the space station (navigating the station menus was slow as all hell though). I've been waiting about five minutes for it to draw the planet now having launched - the screen has frozen with the disc swapping like mad. It seems that it was better before I rebooted so far.
[doublepost=1509943463,1509943291][/doublepost]Aaand it's crashed. pndrun.out just says Killed on the launch shell script line where it invokes oolite --fullscreen I assume. Before that error there's no buffer underruns this time - it's just the normal dmix setup warnings I think, which I always get.

I have 1MB of swap configured currently. Is it worth trying with more?
 

ptitSeb

Serial Porter
Joined
Aug 15, 2012
Messages
8,426
Age
46
Location
France, near Lyon
Killed certainly means it runs out of memory. I wonder how big those OXP are?
Also, I assume you mean 1GB of swap and not 1MB.
 

levi

Still fresh, damnit!
Joined
Oct 6, 2008
Messages
11,373
Location
Somewhere off the coast of the EU
Well, I've removed all the OXPs I can find now, so it should be running with no OXPs.

And duh, good point. I created it with dd bs=1024 count=1024, which indeed is 1 MB, but a swap of one of those versus 512 of the things in my real RAM won't do much. Let me try some more sensible values and report back.

Edit: Yep, a 1GB swap is enough to make the game run a lot more stable. I was able to load a new commander Jameson and hyperspace to Diso, dock, and hyperspace back to Lave. And I was able to load my old commander in galaxy 2, hyperspace to a new planet, shoot down a couple of pirates before crashing in to the planet in frustration that I'd forgotten how to find the damn space station. It certainly seems to need swap to run all of the bells and whistles. If I have the time, I may try to see how far you need to turn the settings down to run it without swap.
 
Last edited:

levi

Still fresh, damnit!
Joined
Oct 6, 2008
Messages
11,373
Location
Somewhere off the coast of the EU
I've just done a fairly exhaustive test using ever decreasing swap, and found that with no swap and settings on Normal Graphics works fine in game. I've experienced one or two crashes after pressing space from after crashing into a station/planet/sun, but no problems at all flying from system to system. I've just done a fairly short delivery run (<10 hops) docking at some stations, and scooping from the sun at others and it's not crashed on the way.

I suggest you should make normal graphics the default. I find the game frustratingly slow especially after hyperspacing (where the hyperspace sample sound and the graphics just don't line up on higher settings) and launching from the space station (normal graphics takes under 5 seconds after launch to display the planet, higher settings require at least twice that). The space stations are perhaps a little hard to see from a distance without the extra effects, but I've not played enough with high settings on to really compare that from many angles - maybe it's just a new feature of the slightly more realistic rendering. You still get the fancy new textures and models from the new version, so it already looks a lot better than it used to with just normal settings.

Leaving shaders and extra an option for people with swap and much more patience than me to try out might be best. And I guess CC users will need swap for sure, if they don't already.

One bug report; I can't seem to get docking approval to work; shift+L seems to be ignored in this version, meaning you either have to have a docking computer, put up with the penalty on an unapproved docking, or turning off docking approval before you go anywhere in the game settings.
 

ptitSeb

Serial Porter
Joined
Aug 15, 2012
Messages
8,426
Age
46
Location
France, near Lyon
Ok, so I have changed the code a bit so the minimum detail is now default when shaders are supported.
Don't forget you can use ES1 backend (by changing appdata/oolite/.es file) if you prefer rendering without shaders.

@levi : about docking clearance, you need to target a station before you can ask for clearance.

Build 19
------------

  • Made minimum detail the default value (still configuration in the pause option screen)
 

levi

Still fresh, damnit!
Joined
Oct 6, 2008
Messages
11,373
Location
Somewhere off the coast of the EU
Thanks, yes docking clearance now works - I must've forgotten how to use it!

I had previously been changing my settings while docked - being able to do it from the pause menu would make testing a lot quicker! However, I've only been able to get out of the settings menu when opened from the pause menu once - the first time.

One copy of my pndrun.out ended like this:
Code:
AL lib: (WW) alSetError: Error generated on context 0x7b1990, code 0xa004
AL lib: (WW) alSetError: Error generated on context 0x7b1990, code 0xa003
AL lib: (WW) alSetError: Error generated on context 0x7b1990, code 0xa004
AL lib: (WW) alSetError: Error generated on context 0x7b1990, code 0xa003
AL lib: (WW) alSetError: Error generated on context 0x7b1990, code 0xa004
AL lib: (WW) alGetError: Querying error state on null context (implicitly 0xa004)
That's it - no FAILED message, none of the normal pnd unmount stuff, and my nubs hadn't been reset. Looks like a hard crash of some sort. Tried with no swap and a small swap, and I got the same.

Sometimes there are less of the Error generated on context... lines - I've seen it once with none, just the final null context error, so they may be a red herring.
 
Top