STK Version Numbers

[charlie]

Since I was derailing another thread, I'll move the discussion here.

I think STK 0.8.2 is not a minor version bump (version numbering conventions tend towards Major dot Minor dot Fix). It represents a point at which a lot of old assets become obselete due to changing requirements, which is always a big hint that a game is evolving into another evolution/game.

Also I think perpetual pre-1.0 development doesn't make much sense. STK long since passed the '1.0' milestone by 95% of software standards. It is played by thousands of people around the world daily, and has been stable for quite some time.

My suggestion, move on from the legacy versioning. Take on a versioning scheme that represents the actual state of the game/development.

I would re-release the "legacy" version (the last version supporting all the assets on stkaddons.net) as version 1.0. This would give a clean cut off point for players that enjoy the old tracks/karts and want to continue on with that experience. It will continue to be played by people who don't have good computers (which is going to be a huge market for many years as developing countries tool up). It would also give a point at which a motivated person could release any bug fixes specific to the legacy version as 1.0.x as well as a version for which the kart/track contribution requirements are a substantially lower bar. I think this legacy version is probably more important than you guys might realise at this point.

The 0.8.2 version becomes 2.0beta with a view to a full 2.0 release once enough assets are updated. This represents the 'next generation' of STK, one that requires a better PC, one for which track/kart contribution standards are substantially higher.

I note that samuncle thinks that 2.0 should be 'multiplayer' but, since this is soley an improvement as opposed to a disruptive change, I think that (in our Major dot Minor dot Fix version scheme) represents a minor increment (even if it is a "major" feature).

[Tiger]

Also I think perpetual pre-1.0 development doesn't make much sense. STK long since passed the '1.0' milestone by 95% of software standards. It is played by thousands of people around the world daily, and has been stable for quite some time...My suggestion, move on from the legacy versioning. Take on a versioning scheme that represents the actual state of the game/development. I would re-release the "legacy" version (the last version supporting all the assets on stkaddons.net) as version 1.0. This would give a clean cut off point for players that enjoy the old tracks/karts and want to continue on with that experience.

I totally agree with you, because I like a lot of the old tracks. I agree too, about the market for STK comprising people with no good computers (I could say, MOST pc owners around the world). Unfortunately, somebody will say "you can download the previous versions, bla, bla, bla..."

[Tiger]

Oops! Repeated answer...

Anyway, I support your suggestion and, IMO, they would ask the community using a poll for this purpose.

[hiker]

I think STK 0.8.2 is not a minor version bump (version numbering conventions tend towards Major dot Minor dot Fix). It represents a point at which a lot of old assets become obselete due to changing requirements, which is always a big hint that a game is evolving into another evolution/game.

Well, we had that discussion as well - I am still open to discuss this, but some reasons for staying the way it is:

• We have a long history of declaring that version 1.0 will be the polished networked version. So releasing version 1.0 (or 2.0) without networking could disappoint users.
• While we do have a new engine, we actually have only two or three really new tracks - not enough imho to call this a new version.
• We had an engine change in the past (from plib to irrlicht), and we didn't make it a major release either.
• Afaik all existing assets should actually still work (i.e. tracks and karts already installed by players). We are mainly moving to a new and improved server which has bug tracking facilities, so we want to get artists to upload their models themselves, so that they can be informed about bugs (instead of people telling us). Just to be precise: 'still work' means there might be some graphical glitches, and of course the existing tracks won't use any of the new features.
• I actually don't mind being having stk being a very conservative version numbering - we don't risk that people say "Version 12.0 is a beta at best".
• I consider the networking to be the much more disruptive change. Yes, with the next version we will potentially leave some old users behind (but we can't do everything - have nice graphics, and support 10+ year old graphics cards - our team is just not that big), but networking (I hope) will be the features that benefits STK most.

Admittedly not very strong reasons One option we are considering is calling the next version 0.9, which then leaves the 0.9.* versions to gradually add networking support.

Also I think perpetual pre-1.0 development doesn't make much sense. STK long since passed the '1.0' milestone by 95% of software standards. It is played by thousands of people around the world daily, and has been stable for quite some time.

Maybe I have very high demands on quality Networking is a very essential feature for a kart racer like STK (imho - and based on the number of requests for networking we get, compared with people complaining about graphics ). In the commercial world I would not release stk the way it is now.

My suggestion, move on from the legacy versioning. Take on a versioning scheme that represents the actual state of the game/development.

I would re-release the "legacy" version (the last version supporting all the assets on stkaddons.net) as version 1.0. This would give a clean cut off point for players that enjoy the old tracks/karts and want to continue on with that experience. It will continue to be played by people who don't have good computers (which is going to be a huge market for many years as developing countries tool up). It would also give a point at which a motivated person could release any bug fixes specific to the legacy version as 1.0.x as well as a version for which the kart/track contribution requirements are a substantially lower bar. I think this legacy version is probably more important than you guys might realise at this point.

It would be great if someone would create an STK-classic version. There was a request to shut down the old addons web site, but (at this stage) we are going to leave it up and running, so that people with 0.8.1 can keep on playing. Having 0.8.1 released as STK-classics would be a good idea. I actually hoped that we could maintain support for older platforms (e.g. by being able to remove unnecessary eye candy from the bigger tracks to keep them playable on older karts), but the graphics developer told me that the overhead is just too much

The 0.8.2 version becomes 2.0beta with a view to a full 2.0 release once enough assets are updated. This represents the 'next generation' of STK, one that requires a better PC, one for which track/kart contribution standards are substantially higher.

I note that samuncle thinks that 2.0 should be 'multiplayer' but, since this is soley an improvement as opposed to a disruptive change, I think that (in our Major dot Minor dot Fix version scheme) represents a minor increment (even if it is a "major" feature).

To make this clear: samuncle is the one in favour of abandoning the existing version scheme My stance is: stk will have 'one chance of fame' (there's no supporting evidence of that statement, it's just my opinion ) - i.e. the one release which will give us enough visibility outside of the open source scene - the 1.0 release. I want to make STK a show case for open source, and to maximise the visibility. So I really don't like announcing "STK 2.0", and then disappointing many players because a rather essential feature is missing. A major version should be a significant change in game play - like a new and improved version, and tbh not much has changed for that - all we have to show atm are 2 (maybe 3) tracks, and options to look better

I am looking forward to other options!

Cheers,
Joerg

[deve]

As I said in other topic, version 1.0 doesn't have sense because it means fully working software (with for example network multiplayer). But there was too many changes to just call it 0.8.2. Break compatibility with add-ons, with older graphics cards etc... I think that 0.9 would be fine. But still it's just a number.

About something-like-STK-Classic... I have a repo https://github.com/deveee/supertuxkart-0.8.1-updates which contains bug fixes for 0.8.1 version. But rather without any feature updates. It is there just for my own use, but if someone is interested in maintaining old version, then some fixes are already ported.

[Akien]

We have a long history of declaring that version 1.0 will be the polished networked version. So releasing version 1.0 (or 2.0) without networking could disappoint users.

I've been following this forum for little less than a year, and I must admit that most of the time I heard "we have a history of saying 1.0 will be the networked version", but I never heard "just wait for STK 1.0, it will have network features!". What I mean is that the impact of changing your milestones would not be so huge as you tend to describe it IMO. I don't know how many people await a networked version that _must_ be labelled STK 1.0 because it was announced as such, but I think it's not that much.

Now, I'm clearly in favour of a more rational versioning scheme (I've also been pushing towards that for OpenDungeons), and IMO STK _is_ a fully working software, as far as open source software goes. It is stable, fun, and is being played by many people all around the world. I don't think aiming for a "commercial-grade" 1.0 version is really meaningful in the free software world, because a commercial 1.0 is often full-feature but buggy software, and you hardly get bug fixes or new features apart from one or two patches before the project is abandoned. Free software is way more awesome: you get a 1.0 version which is already great, and you get promises of bug fixes and awesome new features for as long as the devs find it fun (and you can help too!).

This said, I completely understand the reasons for not wanting to switch from a slowly increasing version number to a "let's bump the major number every two releases".
I think that renaming the current 0.8.2 milestone to 0.9 could be a good start, and then use the 0.9 branch to continue improve the graphics quality (port old tracks, add new antartica-friendly tracks, etc.). Once the assets and tracks are deemed of high-enough quality, you could probably release STK 1.0 as this awesome single- and local multiplayer game (and even if many ask for networking, I do believe that the essence of karting games is local multiplayer; arcade games are way funnier with friends on a couch - on a network, it's more competition-focused and I'm not sure if STK would be that fun as a skill-based competition game).

Then yes, depending on how disruptive the networking feature would be, you could either go with a minor number change or a 2.0 version. In the media culture they love to use "2.0" to denote "connected" stuff, so STK 2.0 could work well in this regard ;-)

[Arthur]

Unfortunately, somebody will say "you can download the previous versions, bla, bla, bla..."

Somebody will say this, because it's true. Just recently one of our previous GSoC students downloaded and played STK 0.3, and it still works fine. So what's the issue?

[onpon4]

Honestly, I don't think SuperTuxKart is deserving of the version number 1.0. Yes, it's fun, but it's also quite unpolished in several areas.

[samuncle]

After speaking with Hiker in private, a good solution would be:

Rename 0.8.2 into 0.9 (+ explaining why we are doing that)
Reserve 1.0 for next gen tracks + network

And IMHO it's the best proposal.

[Akien]

After speaking with Hiker in private, a good solution would be:

Rename 0.8.2 into 0.9 (+ explaining why we are doing that)
Reserve 1.0 for next gen tracks + network

And IMHO it's the best proposal.

Sounds good :-)

[samuncle]

I totally agree with you, because I like a lot of the old tracks. I agree too, about the market for STK comprising people with no good computers (I could say, MOST pc owners around the world). Unfortunately, somebody will say "you can download the previous versions, bla, bla, bla..."

You only see part of the problem.
1) Making an engine compatible with multiple GPU takes a loooooooot of time
2) We are a small team
3) And yes people can download older version. Many commercial games don't allow that (or you have to buy a license). All old versions are available and you play if you want.
4) Rather than criticizing us ^^. Why not contributing for instance by adapting one of the older version to be able to read and play new tracks ? or by removing details in blender

There was a request to shut down the old addons web site, but (at this stage) we are going to leave it up and running, so that people with 0.8.1 can keep on playing. Having 0.8.1 released as STK-classics would be a good idea

I suggested to change the add-on website but not to shut it down . We should make it read-only, so people with older version can still download old tracks but if they want to submit a new track they should use the new website.
Leaving the old addon website open means we have to maintain + review tracks + it can be confusing for people. It's just easier to leave it read only
/!\ That doesn't remove the possibility for people to make their own tracks and share them in their own website.

[Tiger]

Just recently one of our previous GSoC students downloaded and played STK 0.3, and it still works fine. So what's the issue?

None if you don't shut down the old add-on site.

[Tiger]

4) Rather than criticizing us ^^. Why not contributing for instance by adapting one of the older version to be able to read and play new tracks ? or by removing details in blender

I help the project promoting the game among friends and using websites, foruns. Ok! How can I do that? Do I need to be a 3D modeler?

[vlj]

It will continue to be played by people who don't have good computers (which is going to be a huge market for many years as developing countries tool up).

I don't know honestly where this idea come from but I think that in developing regions people have on average newer PC than in developed regions. Not necessarily faster or more powerful though, but newer and thus with GL 3/4 support.
For instance Africa has the highest growth in mobile phone market despite having no "static" phone market. This phenomen is often refered as "leapfrog" and I suspect it also concerns computers. Laptops became affordable quite recently with the netbook appearance in circa 2008.

[GeekPenguinBR]

I don't know honestly where this idea come from but I think that in developing regions people have on average newer PC than in developed regions. Not necessarily faster or more powerful though, but newer and thus with GL 3/4 support.

Me too. I'm from Brazil (wrongly still called developing country for some people) and we are the number two in terms of online population. Laptops are very popular,as well smartphones (note that these stuffs doesn't cost as cheap as in USA), however, not all have PC/laptops optimized for games. So, I agree that many people won't run a "heavier" game easily.

[deve]

In Poland I see a lot of computers with for example intel 965 (gen 4), radeon 9550/9600, sometimes ironlake CPUs with intel HD or at best geforce 310m... If someone bought computer 2-3 years ago, he probably has intel HD 3000 or 4000. These people are simply not gamers nor IT students.

Of course gamers and IT students have high-end nvidia graphics cards But they are playing more interesting commercial games...

I said once or twice that good opengl 2.1 support would be great. This would mean good support for most graphics cards. But if we had limited time, better is to make progress than try to support every possible hardware.

[charlie]

It will continue to be played by people who don't have good computers (which is going to be a huge market for many years as developing countries tool up).

I don't know honestly where this idea come from but I think that in developing regions people have on average newer PC than in developed regions.

Newer != faster. I'm sure there are people in every country with fast/slow/middle PCs and new phones etc. It is an economic reality that there will be many, many more people with under powered PCs (even if they are new e.g. Raspberry Pis, OLPCs) because in general they earn much less per month (by a significant factor).

[Tiger]

In Brazil, laptops with Intel i3/i5 are among the cheapest (equivalent to 600,00 to 1,000 Dollars) and are almost all are out of stock, since the worker class bought all. Afaik, Intel i3 or i5 runs - easily - games that need much more graphic capacity tha STK. BTW: My comment about less powerful computers didn't include Brazil, but other areas in the Eastern hemisphere.

[Anon]

i3/i5 doesn't matter (much) as far as graphics go; that's where the GPU comes into play. And unless they're specifically marketed for gaming, most laptops won't have any kind of a good GPU.

Personally, I think it's probably better for STK to focus on newer technologies over old ones though, since Moore's law will eventually make the newer ones accessible to more people but the older ones will only become more and more obsolete. It's better to prepare for the future, I think, than to focus on supporting older standards.

However, it might also be good to offer a (minimally) updated version of 0.8.1 for people with slower computers. Early on in the development of my temple arena, I ran a test where I made all the library objects in the track local, thereby making it compatible with the 0.8.1 exporters and game. It ran pretty well on a weak HTPC (Athlon II X2 270, geForce GT 610) IIRC. So maybe an updated assets package could be offered for players whose computers can't run the newer versions? It could be presented as "Supertuxkart Classic" or something.

Actually, now that I think about it, the new version does have a fallback mode for when the computer can't the new features. What's wrong with that?

[xeno74]

FYI:

The Windows 2000 version of STK 0.8 has 2219 downloads. And the Mac OS X 10.4 Tiger version of STK 0.8 has 2176 downloads. And lastly the 0.8.1 version for old Linux distributions has 12438 downloads.

[charlie]

FYI:

The Windows 2000 version of STK 0.8 has 2219 downloads. And the Mac OS X 10.4 Tiger version of STK 0.8 has 2176 downloads. And lastly the 0.8.1 version for old Linux distributions has 12438 downloads.

Those numbers do not include distribution downloads (which will likely exceed all of those). I have only ever installed STK from my distro.

[charlie]

About something-like-STK-Classic... I have a repo https://github.com/deveee/supertuxkart-0.8.1-updates which contains bug fixes for 0.8.1 version. But rather without any feature updates. It is there just for my own use, but if someone is interested in maintaining old version, then some fixes are already ported.

I think this needs to not be forgotten.

[onpon4]

i3/i5 doesn't matter (much) as far as graphics go

That's not true. One important factor makes this detail matter: all modern Intel CPUs come with an Intel integrated GPU. Intel also has a habit of pairing more powerful integrated GPUs with more powerful CPUs.

And Intel integrated GPUs have quite a bad reputation, but they're actually perfectly decent. The integrated GPU that comes with my i3 processor seems to have no trouble handling Xonotic at its highest settings. The only reason I can't play Xonotic at its highest settings is the cooling system of my laptop can't keep up. To be fair, I don't know how expensive Xonotic is on the GPU, or how it compares to the beta of STK.

[Akien]

About something-like-STK-Classic... I have a repo https://github.com/deveee/supertuxkart-0.8.1-updates which contains bug fixes for 0.8.1 version. But rather without any feature updates. It is there just for my own use, but if someone is interested in maintaining old version, then some fixes are already ported.

As a distro STK maintainer, I would have love to find about this repo earlier :-)
I'll see if there are fixes worth of preparing an update to Mageia 4's STK 0.8.1.

I also like the idea of a STK-Classic for people with low-end hardware, while the (small) STK team can continue to focus on the new shiny version. I also know of some players who tend to prefer low graphics but awesome gameplay over something too fancy but where it's harder to see the track because of the graphical details (e.g. Crash Team Racing vs Mario Kart 8).

[Anon]

i3/i5 doesn't matter (much) as far as graphics go

That's not true. One important factor makes this detail matter: all modern Intel CPUs come with an Intel integrated GPU. Intel also has a habit of pairing more powerful integrated GPUs with more powerful CPUs.

And Intel integrated GPUs have quite a bad reputation, but they're actually perfectly decent. The integrated GPU that comes with my i3 processor seems to have no trouble handling Xonotic at its highest settings. The only reason I can't play Xonotic at its highest settings is the cooling system of my laptop can't keep up. To be fair, I don't know how expensive Xonotic is on the GPU, or how it compares to the beta of STK.

They still can't really compare with a dedicated video card though. There's a Xonotic performance benchmark, and the Intel HD gpu's all seem to score below 50 FPS, compared with with gaming cards that frequently reach 200+.

From that benchmark, I'd also say that Xonotic is a bit lighter weight than STK. The geForce 660 and 670 series cards all seem to get around or above 200 FPS in Xonotic, while in STK my 650TI and 760 were/are capped at 16 and 31 FPS respectively.

I guess there's an argument to be made that the Intel GPU's will still help with running on lower settings. But if you're going to play without shadows anyway, then you'll probably be hitting a decent framerate with any GPU from the past five years.