Port Arcen games to Linux?

[Pepisolo]

I know this issue has probably been discussed in past, but maybe it might be worth raising again. Let me be brief:

AIWar(+DLC), Tidalis, Valley1+2, Shattered Haven, and Skyward Collapse(+DLC...)

That would make a pretty tasty Humble Indie Bundle, especially when you throw in all Pablo's soundtracks. Also, as far as I am aware, a lot of the games use the same engine, which presumably should make porting for each title easier?

A bit of a long-shot, porting your games to Linux on the off-chance the bundle guys choose your games for a package, but of course there are other reasons to port. Increased sales, especially with Steam on Linux on the rise and the upcoming possibility of a Linux based Steambox hitting homes(?). Could make it a bit of a media event, too, bumping sales for a period. Extra nerd street-cred is always nice.

Am I dreaming or is this an actual (feasible)possibility? Seems like now might be the time to take this gamble. Disclaimer: zero knowledge of unity engine or Linux porting in general.

[keith.lamothe]

If we can get the latest version of Unity to work for our games then the linux port is basically done.  But we've run into serious load-time issues with getting Valley2 on the latest version, etc.

[Pepisolo]

If we can get the latest version of Unity to work for our games then the linux port is basically done.  But we've run into serious load-time issues with getting Valley2 on the latest version, etc.

As simple as that? Update to latest unity, and all games work? Sounds a lot more feasible than I originally thought then, although good luck with all those load-time issues. This might sound a little curt, but isn't this a no-brainer then -- says he who doesn't have to do the programming.

[keith.lamothe]

The problem is that fixing the load time issues is outside the scope of what we can do: the literal filesystem actions individually take over 3 seconds.  No idea why, but we can't fix it.

Nor can we generate a simple case (i.e. not an entire game project) that reproduces it.  That we probably can eventually isolate, but we really don't have time to wrestle with this and whatever other new engine issues may come with the newest version, just for the chance at the additional revenue you refer to.  Just continuing to, y'know, develop games with that time is probably much more likely to help the revenue situation   Skyward Collapse, for example, is really doing pretty well (not record-setting for us, but quite solid).

I hope we do have that time at some point in the relatively near future to get the unity thing sorted out, though

[Pepisolo]

I hope we do have that time at some point in the relatively near future to get the unity thing sorted out, though.

OK. Once sorted it might prove more beneficial than you imagine -- or it might not! Hey, this game development business is a risky business. That HumbleBundle carrot sure looks tasty, though.

[keith.lamothe]

We'd certainly love to get stuff on HB, for sure

And we'd be happy to provide more official linux support in general.  Right now I think all our stuff runs great in WINE, though.

[Echo35]

And we'd be happy to provide more official linux support in general.  Right now I think all our stuff runs great in WINE, though.

It does. I seem to remember the load times being SLIGHTLY longer in Wine, but all Arcen games I've tried ran perfectly fine on it.

[Pepisolo]

And we'd be happy to provide more official linux support in general.  Right now I think all our stuff runs great in WINE, though.

It does. I seem to remember the load times being SLIGHTLY longer in Wine, but all Arcen games I've tried ran perfectly fine on it.

Sounds like in terms of user support, you're at least unofficially fully covered then. It's a shame that about those load-issues though. If it wasn't for that then your name would already be in the HumbleBundle hat, with AIwar +DLC surely being a prime candidate for inclusion.

Oh, looked into this a bit and apparently the Steam Linux user base is growing. It's closing in on the Mac -- for whatever that's worth in profitability terms.

Hopefully those load-issues can be resolved some time in as effortless a manner as possible -- sounds like a bit of a headache, though.

[zespri]

If we can get the latest version of Unity to work for our games then the linux port is basically done.  But we've run into serious load-time issues with getting Valley2 on the latest version, etc.

Out of curiosity, is this run-time or development time issue?
Edit: can you point me to a link or anything that describes the problem? I'm sure you are not the first ones to hit it so it MUST be talked about somewhere on interwebs and I'd like to read it =)

[keith.lamothe]

Out of curiosity, is this run-time or development time issue?

Run-time.  Loading the Unity4 version of Valley2 on some machines (not others) would take minutes where the Unity3 version on the same machine took seconds.  All clearly isolated to the load time of png files off disk during runtime.

Edit: can you point me to a link or anything that describes the problem? I'm sure you are not the first ones to hit it so it MUST be talked about somewhere on interwebs and I'd like to read it =)

There may be some stuff out there on it now but I didn't see anyone else reporting it at the time.  We're somewhat uncommon in our loading of graphical assets at runtime as opposed to just baking them into the asset bundle.  Baking them in would probably work but we'd be losing too much through that change when the only really big benefit would be official linux support:
- development speed would go down due to a lot of re-baking (every time an image was added or changed) and team artists not being able to test stuff out in-game without having Chris or I do a re-bake; or us having to pay a few thousand dollars just for the artists to be able to do that themselves and quite a bit more slowly than they can now
- updates would be hundreds of MB for players to download (and us to upload) instead of 2 to 4 MB because the asset bundle would hold all the images and changing any image would mean including the whole bundle in the update
- players would no longer be able to change any of the game graphics (not a top priority for us, but we won't throw that away just because of some engine bug)

[zespri]

Is it WWW.texture access that is actually this slow, or something else?

[x4000]

Is it WWW.texture access that is actually this slow, or something else?

Sort of. It's angling for that, yeah; but actually the 3-second stall out happens before that, on the www call itself.

Even on some machines where it was not egregiously slower, it was slower to some degree. All very strange.

Since then, Keith has done some Linux/windows/OSX test builds of ai war in a later version of unity 4, and a host of other problems that arose in all OSes based on just upgrading to unity 4 from unity 3. I forget the whole list, but they've been catalogued by the small group of players running those test versions. Perhaps in the AI War forum there's a thread? We were going to do that, but not sure if it is there yet.

Anyway, so there are a number of things. We bought our unity 4 pro licenses almost a year ago, but have been unable to make any practical use of them. We've mostly been waiting to see if they fix the loading speed issue on their own (which they MAY have, it's as yet unclear), because their responsiveness to bug reports in the past have been 0 for 2 over three years of waiting since reporting.

[keith.lamothe]

Since then, Keith has done some Linux/windows/OSX test builds of ai war in a later version of unity 4, and a host of other problems that arose in all OSes based on just upgrading to unity 4 from unity 3. I forget the whole list, but they've been catalogued by the small group of players running those test versions. Perhaps in the AI War forum there's a thread? We were going to do that, but not sure if it is there yet.

Actually it's just TechSY730 that's tried the test version, and I haven't started the thread yet, but yes

Anyway, so there are a number of things. We bought our unity 4 pro licenses almost a year ago, but have been unable to make any practical use of them. We've mostly been waiting to see if they fix the loading speed issue on their own (which they MAY have, it's as yet unclear), because their responsiveness to bug reports in the past have been 0 for 2 over three years of waiting since reporting.

Iirc you told me they did get back to us on the load time thing and asked for a slimmed-down project with a reproducible case.

But when we slimmed it down we couldn't reproduce any absurd load times.


I'm making _some_ practical use of the U4 license now since I'm actually using it for development on my desktop, which lets me get better profiling in AIW, but I'm not using it to build AIW for release so that's still 3.x.  On the other hand, when I run it in 4.x I get a lot of unpleasant textbox oddities that make it slow to enter a bunch of cheats when testing.  Maybe I should go try Cyborg's freepie macros for that

[x4000]

Iirc you told me they did get back to us on the load time thing and asked for a slimmed-down project with a reproducible case.

But when we slimmed it down we couldn't reproduce any absurd load times.

Yes, precisely.  However, we could have spent a lot more time gradually paring stuff out and then seeing at what point it randomly stopped showing that behavior.  If I had confidence they would then turn around a fix, that's what I would have done.  They were extremely proactive in responding to our mention of a bug in 2012 as compared to 2010, but I still have a bit of a lack of trust on that score.  And at the time we were still scrambling to get Valley 2 finished, etc, etc.  Always something more pressing, seems.

I'm making _some_ practical use of the U4 license now since I'm actually using it for development on my desktop, which lets me get better profiling in AIW, but I'm not using it to build AIW for release so that's still 3.x.  On the other hand, when I run it in 4.x I get a lot of unpleasant textbox oddities that make it slow to enter a bunch of cheats when testing.  Maybe I should go try Cyborg's freepie macros for that

Nice on the usage there for the profiling.  I think that's worthwhile.  And I'm not sorry we bought it when we did -- we got a nontrivial discount on it, and I'm positive we will be making full use of it at some point.  We can't stay on 3.3 forever, nor do I really want to.

In general Unity really seems to have a much more proactive and effective manner of bugfixing lately, looking at how they are handling stuff through their blog posts from their lead QA folks.  So that gives me particular extra hope that either they will resolve this on their own (if they have not already, which is still unclear), or that once we have time to hunt through this stuff they will resolve it then.

But with all the new textbox funkiness and so forth, that provides yet more disincentive to try to upgrade, since who knows what all more is there, and if it will be something we can fix on our end, and how long that might take, etc.  As you know.

So, yeah.  Unity 3D is still my favorite engine by a huge margin, but nothing is perfect.  I curse it far less than I would curse any other game engine I've ever seen.

[keith.lamothe]

In general Unity really seems to have a much more proactive and effective manner of bugfixing lately, looking at how they are handling stuff through their blog posts from their lead QA folks.  So that gives me particular extra hope that either they will resolve this on their own (if they have not already, which is still unclear), or that once we have time to hunt through this stuff they will resolve it then.

I'm not sure it will get fixed without a reproducible case, though, as our use of WWW to load textures at runtime appears to be relatively rare.  To the extent that I even now I can google for "unity 4 WWW load time" and not see any results that are actually talking about that issue (it's just general load-time issues, mostly on mobile devices, from folks using the baked-assets approach).

But with all the new textbox funkiness and so forth, that provides yet more disincentive to try to upgrade, since who knows what all more is there, and if it will be something we can fix on our end, and how long that might take, etc.  As you know.

Right, I think we're going to need to make our own textbox unless that "new GUI" thing ever materializes.  It's just getting worse with time

So, yeah.  Unity 3D is still my favorite engine by a huge margin, but nothing is perfect.  I curse it far less than I would curse any other game engine I've ever seen.

Haha, yea, it'd be funny to have them put up a testimonial like "When our company switched to Unity3D, we experienced a 40% drop in the number of times a programmer had to be escorted from the building for excessive swearing!"