Quoting: Perkeleen_VittupääI saw a video about ProtonDB and its current issues by Jason Evangelho https://odysee.com/@LinuxForEveryone:9/the-wrong-experience-for-new-linux:f

A very interesting video, thanks. I, personally, don't mind investing time in making my games work, so protonDB is fine for me. I use it to find workarounds mostly. On the other hand, I do see some problems with it, and in general having an alternative to something never hurt. So, best of luck, Liam.

P. S. Is Jason Evangelho from this video is Jason Evangelho ? I do read his articles sometimes...

Hi, yes, we are doing something. No timeline on it, working on it as and when I have time as an experiment. People everywhere I go complain about ProtonDB, so I'm doing our own thing as an experimental alternative. However, their video was incorrect on one thing: we do not use any data from ProtonDB - what we showed them is demo data.

You can see an example with demo data here. Edit: and yes, you should be able to play with it and submit your own demo reports to try it out ;)

I'm also tracking what's left to do / plans and such [over here](https://gitlab.com/liamdawe/gamingonlinux/-/issues/385). Unlike other sites, GOL is 100% open source.

Feel free to let me know what you think. Still entirely unsure about it.

Last edited by GamingOnLinux Bot on 18 Oct 2020 at 3:08 pm UTC

I stopped submitting reports to ProtonDB when they brought in the Steam account-linking requirement. I'm sure I'm not the only one.

Having a better repository of information to point people to would be useful, I think. I hope one day it will end up as just a very short list of games that don't work, and everything else is either native or works without intervention.

Something that would share Wine and Proton reports would be even more useful. Wine AppDB is somewhat archaic at this point especially since it's primarily focused on upstream Wine and at most staging. ProtonDB had a reverse problem - it was for Steam games only.

While I prefer upstream Wine, a lot of people are using Proton, so it's good to have a shared database.

Last edited by Shmerl on 19 Oct 2020 at 2:39 am UTC

Quoting: ShmerlWhile I prefer upstream Wine, a lot of people are using Proton, so it's good to have a shared database.

The issue there would be the same one complained about in the video for ProtonDB, I think: that it would tend towards a bug-tracking and tinkerer's repository.

For Proton, you want to be able to know before you buy the game whether you'll be able to just press the button to play. As I understand it, that's what Liam's going for here.

For upstream Wine, you have that in Lutris: you hit the play button and the runners take care of how that happens.

In addition to those, having a central bug-tracking and tinkerer's repository for submitting reports would be useful, to drive development of Wine/Proton, and to inform the creators of Lutris runners or those that want to do it themselves. That should come from CodeWeavers, really, I'd say, as the people that are involved in both Wine and Proton. That's a different audience than the one this initiative is aimed at, though, and the way people want to be able to use ProtonDB.

Quoting: CatKillerThe issue there would be the same one complained about in the video for ProtonDB, I think: that it would tend towards a bug-tracking and tinkerer's repository.

I think you totally need both. Some games require tweaks to run, and the database showing how to do it is a valuable resource (that's what Wine AppDB is doing). But my point stands for having a resource for all flavors of Wine in one place.

For upstream Wine, you have that in Lutris: you hit the play button and the runners take care of how that happens.

I find solutions like Lutirs to be poorly maintained. As the number of games grows, the quality of the scripts goes down and they fall out of date faster. Same happens with PlayOnLinux. It's a lot more clear where there is a database entry which shows workarounds and also says when the report was made last. So you can figure out it can be stale if it was years ago. With solutions like Lutris it's very obscured.

So I'm personally not using Lutris or POL anymore but simply install games and refer to database reports for needed configuration if something doesn't work.

Last edited by Shmerl on 19 Oct 2020 at 3:07 pm UTC

Our solution is just for Steam+Proton, it's part of the point. Keeping it simple. We can look at perhaps doing other things for native+wine somewhere else but for now our reports are just for Steam+Proton.

If you have feedback, feel free to make new topics on it for me to follow as this is already full of talk about what we're not doing...

Quoting: Liam DaweOur solution is just for Steam+Proton, it's part of the point. Keeping it simple. We can look at perhaps doing other things for native+wine somewhere else but for now our reports are just for Steam+Proton.

Thanks. I think Wine AppDB already covers it (not perfectly but OK) for reguluar Wine. I thought it could be an improvement to have something combined which so far didn't exits.

Quoting: TriasP. S. Is Jason Evangelho from this video is Jason Evangelho ? I do read his articles sometimes...

Yep! It's the same guy!

Quoting: Liam DaweHi, yes, we are doing something. No timeline on it, working on it as and when I have time as an experiment. People everywhere I go complain about ProtonDB, so I'm doing our own thing as an experimental alternative. However, their video was incorrect on one thing: we do not use any data from ProtonDB - what we showed them is demo data.

You can see an example with demo data here. Edit: and yes, you should be able to play with it and submit your own demo reports to try it out ;)

I'm also tracking what's left to do / plans and such [over here](https://gitlab.com/liamdawe/gamingonlinux/-/issues/385). Unlike other sites, GOL is 100% open source.

Feel free to let me know what you think. Still entirely unsure about it.

Already now it looks more useful, coherent and clean in comparison to modern protondb. Nowadays on protondb: one really cannot tell what means what!

I'm sure many would love to see this project taken forth. Should you guys run a crowdfund? Ask ppl to get your merch? Something else? Little bit of "competition" obviously would not harm in this case. I mean, protondb was brilliant at start and then something happened in their approach.

Gold used to mean: a little tinker effort required; launch option or similar and that's it. Now gold seems to mean you must install a custom proton +whatnot to get a "gold" game even launch. It's all in that Linux 4 Everyone video. No. Just no. This is not the way to go much longer. We need simplicity and clarity whether a game works on Linux OOTB or not.

And already are the new comers to Linux gaming in a total state of confusion..

Quoting: Perkeleen_VittupääGold used to mean: a little tinker effort required; launch option or similar and that's it.

See... Tought Gold meant "runs, no tweaks, but might have a tolerable performance hit (15-40% aprrox.). We've seen Linux native ports having that kind of performance hits. While Platinum meant "runs, no tweaks with native-like performances".

Gold with tweaks makes no sense to me. It should systematically fall in the silver grade, no matter the performances.

Last edited by Mohandevir on 19 Oct 2020 at 8:57 pm UTC

Quoting: Mohandevir

Quoting: Perkeleen_VittupääGold used to mean: a little tinker effort required; launch option or similar and that's it.

See... Tought Gold meant "runs without tweaks but might have a tolerable performance hit (15-40% aprrox.). We've seen Linux native ports having that kind of performance hits. While Platinum meant "runs, no tweaks with native-like performances".

Gold with tweaks makes no sense to me.

See? Now no one knows what exactly means what. What gold used to mean to me (and as indicated in that Jason's video) it was a good "GOLD" experience in performance, but only after a little tiny tweak (what ever THAT again means to different individuals). What is a "simple command in the terminal" for us, could be a deal breaker for some one else, new user or so.

Quoting: Perkeleen_Vittupää

Quoting: Mohandevir

Quoting: Perkeleen_VittupääGold used to mean: a little tinker effort required; launch option or similar and that's it.

See... Tought Gold meant "runs without tweaks but might have a tolerable performance hit (15-40% aprrox.). We've seen Linux native ports having that kind of performance hits. While Platinum meant "runs, no tweaks with native-like performances".

Gold with tweaks makes no sense to me.

See? Now no one knows what exactly means what. What gold used to mean to me (and as indicated in that Jason's video) it was a good "GOLD" experience in performance, but only after a little tiny tweak (what ever THAT again means to different individuals). What is a "simple command in the terminal" for us, could be a deal breaker for some one else, new user or so.

What I meant by "See" it's that, aside from Platinum, the ProtonDB ratings are just confusing. It gets even worse when you try to find the mentionned tweak everyone talks about which was posted 125 reports earlier. I have a lot more confidence in what Liam (GoL?) is trying to do. I believe it would be a lot more reliable.

Last edited by Mohandevir on 19 Oct 2020 at 9:05 pm UTC

Quoting: MohandevirIt gets even worse when you try to find the mentionned tweak everyone talks about which was posted 125 reports earlier. I have a lot more confidence in what Liam is trying to do. I believe it would be a lot more reliable.

This so much! One of the worst parts in the modern protondb of a mess is that one cannot find the info needed. The requirement to even realize that one has to search for some launch option strings (and where exactly copy paste that..) in midst of a wall of reports is insane.

Why not a box with "marked as solutions" at least somewhere on the side or similar? All comes down to UX by the design skill. Apart from coders, we need some people to help who can place themselves in the shoes of a new Linux user, trying to get a game one loves to run with as less effort as possible.

It really is good that this is on the table now the very issue, at last, to discuss. It can only get better :happy:

Quoting: Perkeleen_VittupääAlready now it looks more useful, coherent and clean in comparison to modern protondb. Nowadays on protondb: one really cannot tell what means what!

The thing with ProtonDB is that its state is a consequence of how it came to be.

When Proton was announced, it came with a tiny whitelist of games that Valve pledged to support and which were listed on Steam as Linux-compatible, as well as the option of trying it with other games. So everybody did, of course, and put their results in every thread where Proton was announced. Interesting, but in no way comprehensive or discoverable.

Someone then collated those results they could find into a big spreadsheet. That got unwieldy pretty quickly.

So, ProtonDB was made, and it used the same medal rankings as Wine. Unfortunately, in their exuberance that things were working, reporters weren't really following those categories in their reports. If it worked at all people were calling it platinum because they were excited, and wanted to see the number of platinums go up.

So ProtonDB wanted to clean up the quality of their reports. While that was a noble goal, the two big problems I see with the approach that they took was that firstly they insisted on linking reports to a Steam account, which drove away the filthy casuals like me and selected for enthusiasts and tinkerers instead, and secondly, rather than enforcing reports into the medal categories they did away with them on reports, but still display those rankings as some sort of average, which makes no sense.

At the same time, we do, as a community, need a simple way for new migrants to know which of their games are likely to work, and they get pointed at ProtonDB because we don't have anything else (currently), even though it's not that well-suited to that role.

Quoting: CatKiller

Quoting: Perkeleen_VittupääAlready now it looks more useful, coherent and clean in comparison to modern protondb. Nowadays on protondb: one really cannot tell what means what!

The thing with ProtonDB is that its state is a consequence of how it came to be.

When Proton was announced, it came with a tiny whitelist of games that Valve pledged to support and which were listed on Steam as Linux-compatible, as well as the option of trying it with other games. So everybody did, of course, and put their results in every thread where Proton was announced. Interesting, but in no way comprehensive or discoverable.

Someone then collated those results they could find into a big spreadsheet. That got unwieldy pretty quickly.

So, ProtonDB was made, and it used the same medal rankings as Wine. Unfortunately, in their exuberance that things were working, reporters weren't really following those categories in their reports. If it worked at all people were calling it platinum because they were excited, and wanted to see the number of platinums go up.

So ProtonDB wanted to clean up the quality of their reports. While that was a noble goal, the two big problems I see with the approach that they took was that firstly they insisted on linking reports to a Steam account, which drove away the filthy casuals like me and selected for enthusiasts and tinkerers instead, and secondly, rather than enforcing reports into the medal categories they did away with them on reports, but still display those rankings as some sort of average, which makes no sense.

At the same time, we do, as a community, need a simple way for new migrants to know which of their games are likely to work, and they get pointed at ProtonDB because we don't have anything else (currently), even though it's not that well-suited to that role.

Very insightful post! "Platinum because they were excited" - and so forth indeed. Oh yea.

These are the reasons why i think rational Linux people (like LiamD and Samsai for example) of GOL (as this project is indeed under GOL) would be effective to CURATE stuff. Listening to their recent podcasts, they're a great team with needed fiber to "swing the axe" as necessary to keep this new "protondb" in shape and reliable!

"Blockchain... more like blockfail." -not many dare to state even that in these times of many changes in general. "Not all change is progress" (as the long gone Linux podcast Linux Luddites once stated) :smile:

Quoting: MohandevirSee... Tought Gold meant "runs, no tweaks, but might have a tolerable performance hit (15-40% aprrox.). We've seen Linux native ports having that kind of performance hits. While Platinum meant "runs, no tweaks with native-like performances".

Gold with tweaks makes no sense to me. It should systematically fall in the silver grade, no matter the performances.

For the record, your understanding was incorrect. If a game has performance issues that can't be corrected, it's Silver if the performance is adequate or Bronze if it's not. Gold and Platinum would both be running without any issues, performance or otherwise, but differing in whether you need to do anything to make that happen. I'll list the criteria.

Quoting: AppDBPlatinum
Works as well as (or better than) on Windows out of the box.

Gold: Works as well as (or better than) on Windows with workarounds.

Silver: Works excellently for normal use, but has some problems for which there are no workarounds.

Bronze: Works, but has some problems for normal use.

Garbage: Problems are severe enough that it cannot be used for the purpose it was designed for.

Quoting: ProtondbPlatinum (runs perfectly out of the box)

Gold (runs perfectly after tweaks)

Silver (runs with minor issues, but generally is playable)

Bronze (runs, but often crashes or has issues preventing from playing comfortably)

Borked (game either won’t start or is crucially unplayable)

Quoting: CatKiller

Quoting: MohandevirSee... Tought Gold meant "runs, no tweaks, but might have a tolerable performance hit (15-40% aprrox.). We've seen Linux native ports having that kind of performance hits. While Platinum meant "runs, no tweaks with native-like performances".

Gold with tweaks makes no sense to me. It should systematically fall in the silver grade, no matter the performances.

For the record, your understanding was incorrect. If a game has performance issues that can't be corrected, it's Silver if the performance is adequate or Bronze if it's not. Gold and Platinum would both be running without any issues, performance or otherwise, but differing in whether you need to do anything to make that happen. I'll list the criteria.

Quoting: AppDBPlatinum
Works as well as (or better than) on Windows out of the box.

Gold: Works as well as (or better than) on Windows with workarounds.

Silver: Works excellently for normal use, but has some problems for which there are no workarounds.

Bronze: Works, but has some problems for normal use.

Garbage: Problems are severe enough that it cannot be used for the purpose it was designed for.

Quoting: ProtondbPlatinum (runs perfectly out of the box)

Gold (runs perfectly after tweaks)

Silver (runs with minor issues, but generally is playable)

Bronze (runs, but often crashes or has issues preventing from playing comfortably)

Borked (game either won’t start or is crucially unplayable)

Yeah, but I remember that the classification changed since Protondb began collecting reports. I can't find the original version, but it wasn't categorized in the same way, originally. They did a complete refactoring of the creterias, at some point, last year.

Anyway, it doesn't really matter. To me, ProtonDB is starting to be unusable/unreliable. It is an awesome initiative/starting point, but it's flaws are getting more obvious by the day.

Edit: I'm wondering if we'll ever see Valve expending it's official Whitelist?

Last edited by Mohandevir on 20 Oct 2020 at 4:18 pm UTC

Quoting: MohandevirYeah, but I remember that the classification changed since Protondb began collecting reports. I can't find the original version, but it wasn't categorized in the same way, originally. They did a complete refactoring of the creterias, at some point, last year.

They started with the classifications from the spreadsheet (Won’t Start, Crashed, Unplayable, Unstable, Stable, and Completely Stable) and then changed them to the Wine ratings about two years ago, and then to their current thing about a year ago.

Edit: I'm wondering if we'll ever see Valve expending it's official Whitelist?

They've expanded it a couple of times since the initial release. You can see which games are on it [here](https://steamdb.info/app/891390/). Since whitelist inclusion means that Valve have to shoulder the support burden rather than the game devs (which they did as an incentive in case games needed changing to work well with Proton) and an amazing number of games work without intervention, they seem to have decided to let the game devs handle support and just make Proton better themselves.

Quoting: CatKiller... and an amazing number of games work without intervention, they seem to have decided to let the game devs handle support and just make Proton better themselves.

My feeling too.

Still, what's next for the whitelist? The last update was for Proton 4.2 games... It's been quite a while... I get the point that testing all the games is a massive undertaking, but isn't there anything else that Valve will/could do to automatically integrate more of the "Platinum rating games"? I don't know, maybe a community of reliable and approved testers, like Liam and GoL? :wink:

I'm sure Valve has no intentions of relying on ProtonDB reports for that kind of stuff, though.