Insane Difficulty: A Request for all SNES Modders and Beta Testers - Insane Difficulty

I meant to post this sooner, but if you're planning to develop or beta test any SNES mods, this is probably worth a read.

Please use accurate emulators for all testing and development.

Accurate emulators are:

SNES9x (not 1.43), use 1.51+ to be safe
Higan/Bsnes

Please don't use ZSNES for testing or reporting bugs or developing a mod. Why?

ZSNES is old and out of development. It hasn't received an update in nearly a decade. On top of that, its original design goal was simply to work, often with the most popular games being hacked in to work, not to accurately portray how a game should play. If a game isn't part of that group of popular games, it will be a mess to try and make it work with ZSNES. Even if it is, developing around an inaccurate and ancient emulator is a poor design choice. It would be roughly the equivalent of writing software for windows 95.

Why is accurate emulation important for testing and development?

In an era of Flash Carts and Repro Carts, playing a mod on actual hardware is a reality. Neither SNES9x nor ZSNES accurately emulate the hardware cycles of the SNES. In both cases, those programs provide a workaround to avoid the massive processing demands required to do it properly (which is what Higan strives to do). In doing so, situations arise where things seemingly behave perfectly but when played on console they don't actually work. This exact thing happened with BNW's Esper Bank and had I not discovered it (and thanks to MechanicalPen, learned exactly what it was), the people making the repro carts for BNW would have had a broken game on their hands. Similarly Zombero's Ogre Battle hack doesn't behave properly on hardware. I suspect a similar issue is the cause.

My computer can't run Higan/Bsnes!

Use Higan-Performance, if possible, otherwise use SNES9x. SNES9x is very light weight and should have no issues running on a system that can run ZSNES. It's fine for just about every purpose and is still actively developed. But at some point, a mod needs to be tested in Higan to see how it would perform on original hardware. Higan isn't perfect, but it's very close.

Mish, you're just a purist and you need to get off your high horse.

Maybe. But I'll counter that by saying that it would be wise to remove an attachment to ZSNES from the needs of a modding community's beta testing and development. Play a game in whatever way is enjoyable, but develop and test a game in a way that is the most accurate.

For more in depth information, I would encourage the reading of Biyuu's article on emulation accuracy. I pulled a few of my statements from it.

http://arstechnica.c...-snes-emulator/

Designing a game to work only on ZSNES is a mistake, and you're right about that. But your misconception is that games shouldn't be tested in ZSNES whatsoever.

You can make a million threads and post a thousand links showcasing that ZSNES is objectively worse than snes9x and higan, and you will be right. But I want my game to be played, and games nowadays are played in both snes9x and zsnes. And that won't be changing any day soon.

I'd be making a mistake by testing in only one of the two emulators. So my advice is that you test your games first in snes9x, and second in zsnes.

Yo, Mish, I know this is near and dear to you so I took the liberty of making a diagram.

I hope you like it.

I guess the TL;DR is get snes9x if in doubt. So that diagram is appropriate, Arc.

Re: Hart.

I'm not trying to tell someone that playing casually with ZSNES is wrong. My point is that striving to make a game work the same on all emulators isn't realistic because they don't all emulate the same. If they don't emulate the same, it stops being the same game. If people report bugs in ZSNES that aren't duplicated in snes9x or Higan, it's not good information.

If it works on snes9x, you're probably good to go for most cases. If you want it on a repro cart or you think people will put it on hardware, it better work on Higan.

Snes9x is so light weight and easy to use, it should be a very viable replacement for ZSNES in all cases.

Mishrak, on 15 March 2016 - 08:17 AM, said:

I am. zSNES is fucking trash. Get a real emulator.

Stop playing the game wrong.

Brave New World was largely developed on ZSNES >.>

10 years ago, when I chose it over SNES9x, it was the superior choice. It... simply no longer is.

Mishrak, on 15 March 2016 - 08:17 AM, said:

Unless you want that part of your audience that plays with ZSNES to have a bug-free experience.

That's taken a bit out of context, Stann, but it's still a valid point. My post is geared towards accurate emulation. If accuracy isn't really a concern, then this thread is a non-issue. If someone is making a repro cart, or there's an expectation that the mod will be run on original hardware (ala MMX HT), accuracy should be the utmost importance. If it hasn't been tested properly, and it doesn't work on hardware, it cannot be remedied by using ZSNES.

If the bugs in gameplay aren't caused by poor emulation, then yes ZSNES feedback would be good. But if the issue isn't duplicated in SNES9x and Higan and is localized in ZSNES because of its shortcomings, how could a developer be expected to make a bugfix when there's an equally lightweight emulator where the bug doesn't occur? At that point the developer would be "fixing" something that is actually already working properly.

Let's just say that when I visit a website in Firefox, and it tells me "You seem to be using an outdated browser. This website has been developed for Chrome." I immediately back out and don't visit that website. I prefer to use Firefox, just as I prefer to use ZSNES. It's up to the developer if they want those of us who prefer ZSNES to play it, and if they do then our feedback is just as useful. If not, then whatever.

Stann, on 15 March 2016 - 10:45 PM, said:

That's not too relevant to the thread, however. His points are completely valid in context. Other than that, you're not at all wrong. I'm pretty sure that regulars here know my approach to software in general. "meeeeh, it works and I like it" rarely comes into the equation when I discuss software, as that would render all discussions moot for eternity.

I'm not saying that what you're saying is irrelevant in general, just in this particular case. Appeasing the user base comes later.

BTB, on 15 March 2016 - 01:44 PM, said:

O.O

There is still hope for you/For once, you're speaking my language.

I'm pretty sure it's relevant. If a modder wants their mod to be played by the largest audience possible, they should keep in mind that many people still prefer ZSNES over Snes9x. If that's the case, feedback from those people is just as important. If that's not the case, then their feedback is irrelevant. The same thing can be said about website developers; not everybody likes Chrome, so making your website only usable with Chrome alienates many of your potential customers.

Mish proposes an argument that's focused around the way he likes to play. I just think it's a stretch to over-emphasize how important it is to a majority of us players. If your mod runs perfectly on native hardware, but doesn't on someone's preferred emulator of choice, that person is probably not going to play it. And it's up to the modder who they want to accommodate, but just saying "Don't give feedback if you're using ZSNES because some people like to play on Repro carts." is kind of missing the point.

BNW runs fine on ZSNES, Snes9x, the shitty emulator I have on my PSP, and thanks to Mish's feedback on his native rig. While it may take extra effort, it's clearly not that difficult to be accommodating, and the extra work is very much appreciated by those of us who want it.

I guess we look at "beta test" a bit differently? That sounds like an active choice with intent. Reporting a bug you find while you're just playing the game is to me more of a "good samaritan" thing. I wouldn't label everyone reporting a bug as a "tester". Hence why I feel like the focus should be incremental instead of having an all-at-once approach. First make it work like you want it on an accurate platform, ironing out for other platforms (emulators, in this case) comes later on.

I think we agree with each other in the end, and perhaps only disagree on how (and when) to get there.

ETA: ...I do agree that "Don't give feedback if you're using ZSNES because some people like to play on Repro carts." is a pretty poor stance as a whole, though.

Kaffe Myers, on 16 March 2016 - 02:30 AM, said:

This is fine, assuming that the modder wants to go down that route. And if they do, great. If not, also great (For the rest of us who don't care about accurate emulation.)

I'm admittedly not very good at reading today (attention deficiency galore), so I'm unsure on how valid my responses are. That being said...

Stann, you're so damn easy going. xD
"If people wanna fuck side-ways, then let them fuck side-ways!"

And I'm like "THAT'S HOW YOU GET HOLLOWBORNS!"

But in the end, we kinda aim for the same thing.

Kaffe Myers, on 16 March 2016 - 03:00 AM, said:

Though some of you are aiming sideways, apparently.

It's not just because I prefer to play on hardware. It's also that I want the game to behave correctly. The case I'm trying to make for that is that a developer will be hard pressed to make their game work across all platforms. The browser analogy fails a bit because Chrome/Firefox are generally equal at allowing you to browse. (Screw IE) ZSNES is not an equal emulator to snes9x. Maybe on the surface, but certainly not from a technical standpoint.

The entire reason I suggest that ZSNES isn't good for bug testing is it's not a good emulator. I'm saying that as objectively as I can - meaning, it doesn't emulate games well. It's not about what people prefer to use, it's about whether or not the game is working correctly. In that regard, ZSNES is not even in the ballpark. That it works half decently is almost sheer luck. If a game works well on snes9x it's basically fine on hardware in all but the finest technical situations. The same cannot be said of ZSNES.

I'll use the same example in the Ars article, namely:
https://www.youtube....h?v=jWZ7Q6U2x-c

There's nothing a developer could do to make that game behave correctly on ZSNES because the problem is with ZSNES.

It seems like my point wasn't heard.

We modders can't overlook the fact that ZSNES is one of the most used ways to play SNES games, if not the most used. That's a fact, whether you like it or not. And you can try to convince us for a decade that ZSNES is a bad emulator, and you might even convince us. We (with "we" I mean people who think like me, who want as uch people playing the mod as possible) will still try our best to support ZSNES (as well as native hardware).

I've told you in chat yesterday that your point would come across a lot better if you focused more on highlighting how good SNES9x is as an emulator, instead of talking ZSNES down - Because your objective is for mods to be supported on native platforms, and not to bash against a single emulator. You really sound like you are just doing the latter right now.

You're missing my point as well so this really isn't much of a discussion anymore.

The emulators are not equals and just a matter of preference. One is objectively not good at emulating. It's not an opinion.

Changing a mod just to work on it, when it already works fine on other emulators is poor design philosophy imo whether the focus is hardware performance or not.

Reporting bugs in an emulator that are localized to that emulator because it's not good at emulating is bad information because it's not the game's fault at that point.

The problem is that even though SNES9x is objectively better, I don't think anyone is arguing that, there's still a wide player base that uses ZSNES. If you don't take that into account you start to alienate some of your player base.

I could be wrong but from what I've seen, most modders seem to be more interested in making their mod playable for everyone instead of guaranteeing a higher emulation accuracy. If someone starts playing on ZSNES without knowing any better and it ends up crashing over hour in then a fair amount of people might just give up on it rather than figure out it was because of the emulator and start over.

Things like that mean that less people would be likely to share/recommend it and that fewer people could enjoy it overall. I know when I first started getting into emulating games I sure as hell didn't know the difference between one emulator or the other and just picked up ZSNES because a friend mentioned it.

So if someone ends up spending slightly more development time, or adding fixes that may not be needed on accurate emulators, and get a mod to work on ZSNES they may consider that worthwhile since it can help to expand the player base and result in a better overall experience for their players. It mainly depends on the goal of the modder.

Great post Tray.

I'm happy to see that a third gets both mine and Mish's points (I consider both valid). It's two different points of view that a lot of modders and players should be aware of when playing and designing mods.