2.00 Aim for cross-platform?

[namida]

Final decision: Due to difficulties in compiling a Linux version that works of even just a small test app, I do not plan to proceed with cross-platform. Sorry.

So, in doing the V2.00n mostly-rewritten code, I'm wondering whether people feel it's worthwhile aiming for cross-platform? I will stress that this won't be a matter of "every system in existance" - I don't at all plan to target Android, or homebrew-capable consoles, for example. I have my doubts about even going for Mac, for the simple reason that I will most likely never own a Mac myself and thus will be unable to test. However, Linux is a realistic possibility; I now have a secondary PC, which runs Ubuntu, so I am able to test a Linux version. If I switch over to Lazarus - something I'd quite like to do regardless, if possible - then it shouldn't be too difficult to also create and support a Linux version.

On the other hand, the benefits of such aren't too huge, since WINE exists.

How many of you's use Linux, and if so, do you think that a native Linux version of NeoLemmix would be a good thing to aim for? I will stress that the Windows version will always remain the primary version.


(If there's much benefit to be had, a native 64-bit Windows version is also a possibility, though I don't see much need for this as the 32-bit version runs fine on 64-bit Windows.)

[IchoTolot]

Well Lemmini ran (a bit surprisingly) on the mac of my old school friend, so maybe NeoLemmix also runs on a mac. I could try it out the next time I meet him and report the results.

I am using Windows 7, but Linux compability would be nice, because I am thinking to get a Linux laptop for programming (well and for a little bit of lemmings in breaks )

[namida]

Lemmini runs through Java, which is cross-platform by design. Other than by rewriting NeoLemmix in Java (which I don't intend to do), it can't be ported the same way. Lazarus, on the other hand, is the same language as NeoLemmix already uses, just a slightly different dialect, and porting any parts I don't rewrite should not be a very complicated task. However, I believe WINE exists for Mac, which could enable running it.

While I could theoretically compile a Mac version (I think) once it's on Lazarus, I'd have no way of testing it, thus at most it'd be available-but-not-officially-supported.

[607]

Hm, I do use linux, but I still don't really see much need for this. I think most people with Linux will either have Windows as well or will have enough knowledge of the system to emulate it (emulate is probably the wrong word, looking at what WINE stands for, but I don't know how I should call it).

[GigaLem]

might as well go for it since there could be people who want to create levels for lemmings and needs the respective os to do so

[namida]

So, based on some quick tests and Google research, it seems like cross-compiling from Windows to Linux is a hell of an ordeal to set up. From what I hear though, the reverse isn't so bad, so the option is there of developing on my Linux machine. Or I could just compile the Windows version on Windows, and the Linux version on Linux... this does rely on the assumption that I'll always have two PCs though (or a dual-boot setup).

[mobius]

I also have an Ubuntu PC which I only ever use when this one breaks down. It would be nice I guess but I personally don't care too much.

Can anyone verify how well it runs on wine? Last time I checked (which was a long time ago) what I needed to run [some audio editing programs] ran surprisingly well on wine, compared to the last time I tried years before that. I'd assume they improved.

[namida]

I haven't tried recently; I could try it on my alternate PC, but given that it's a fairly weak netbook (the whole reason I put Linux on there - Win7 doens't run very well on it, and I don't want to use XP), I don't think it would be a great indicator. I also have my old laptop around, if I can get it to stay turned on for more than 10 minutes I can try chucking Linux on that. For the record, when I tried regular Lemmix on WINE ages ago, there were occasional sound-related issues but music, graphics and gameplay were all fine. At that time, Lemmix used a different audio library for sound than it did for music; NeoLemmix now uses the same one for both (a very close relative of the one it formerly used for music) so will probably have no issues in that regard either.

I believe Simon primarily uses Linux, and he's played at least some NeoLemmix stuff, so he can probably give you a good idea of how well it works.

[Simon]

I'm running a 9-year-old laptop (2006/07) with Debian 6 (released 2011) with mostly security patches, a somewhat outdated Wine, but recent video drivers.

I'm experiencing slowdowns with many lems out. Fast-forward doesn't increase the speed anymore around 20 or 40 lems out. Skipping 10 seconds is fast, so the slowdown comes from the graphics.

I haven't built levels in the editor much, neither on old Lemmix, or Neolemmix.

Starcraft 1 in Wine (1998, DirectX 5) slows down with several units on-screen. It's too slow for a LAN game. It ran normally fast under Windows. Badly configured Wine could be a reason, or suboptimally configured video drivers.

Graphicsgale (pixel art image editor) is slow while scrolling. I use this more often than other programs in Wine.

A native version of Neolemmix might run noticeably faster.

-- Simon

[namida]

After looking some more into this, it appears that if I want to create a Linux version, I'll have to use a different graphics library in place of Graphics32. This should be feasible, but that does raise the question of exactly which one to use...

[geoo]

Obviously a native Linux version would be cool, but I'm not sure if it's really worth the effort of porting it to a different graphics library. Depending on how well written and modular Lemmix is, this could be anything from a reasonable task to a nightmare.

For me, Lemmix runs decently under wine, and I assume the same applies to NeoLemmix unless you did some really weird stuff. So maybe your time is spent better implementing the new features well and getting the release out a bit earlier.

[namida]

It won't need to be "ported" as such, as that side of things is being rewritten anyway. More, it comes down to that I have to learn how to work with a different one; and find a suitable one for that matter. Graphics32 is ideal for pixel-based operations; the biggest downside is lack of support for hardware acceleration (all graphics are handled by the CPU).

[geoo]

SDL 1.2 naturally springs to mind as a natural candidate, well tried and tested. Mindless has used it, NaOH, I've meddled around with it too and major titles from Valve have used it. SDL2 if you want to be modern.
Also handles sound and stuff.

[ccexplore]

You can architect it for cross-platform but without explicitly supporting multiple platforms by making sure to only use libraries that are cross-platform (eg. SDL as geoo suggested), then open-source it to let others with the ability and skill to build and test for the platforms that you don't/can't officially support yourself.  "Officially support" in this sense means the platforms you actually build, test and bug-fix the game for.

So for example, you can just focus building, testing and fixing bugs for, say, Windows.  Because the underlying code and architecture is cross-platform, most of the code updates will likely carry over to other platforms as well.  People using the opened source for other platforms can focus mostly on whatever extra tweaks are needed to get it to work in those platforms you aren't testing yourself.

All that being said, Windows is major enough of a platform that it seems you can find viable virtualization/emulation solutions for most other desktop OSes to deal with a Windows-only version, so like geoo said, maybe the effort is better spent elsewhere.

[namida]

From the looks of things, it appears that if I write a custom class to handle the behind-the-scenes image processing (which doesn't exactly need to be very complex for NeoLemmix; and has advantages in that I can easily expand these to handle things other than just a bitmap; eg. trigger areas), then the built-in Lazarus components may well be sufficient. In this case, no special library for graphics needs to be found.

And, BASS is already multi-platform, so audio library is already handled.

In regards to platforms I'm able to officially support (ie: have access to and can test on), it's just Windows and Linux; unless I also decide to do an Android port at some point (which is unlikely). It might also be possible for me to compile it for Mac, but I have no way of testing such a port.

Overall, it would appear there's some interest in a Linux version. Since developing such would also be good experience for me, and shouldn't entail too much extra difficulty, chances are good that such a port will exist.