Quote from: Simon on August 23, 2026, 09:24:47 PMLemmings 3 happens to implement what you're considering. Solutions from an action replay don't count ("learner mode") for progression. Feature is meticulously implemented, but you're not supposed to use it until you've proven your worth. Reaction by Kieran:

Quote from: kieranmillar on July 17, 2023, 10:54:20 PMusing the replay feature makes the solve not count (why oh why did they do this? Idiots)

They did this because using the replay system to solve a level was probably considered cheating (or at least contradictory to the spirit of the game as it was intended to be played) by the developers. Otherwise, of course they wouldn't have implemented it in this way.

Part of the original challenge of Lemmings was to be able to solve the level in one single playthrough. If you could do that, you'd beaten the level good and proper, fair and square. And, of course you could pause the level to take a moment to set up your next action, but all assignments had to be made in real-time: that was a legitimate and enjoyable part of the challenge. I can't be the only player that thinks like this.

Anyway, RetroLemmini has already sacrificed some of this by having a replay system. So did Windows '95 Lemmings (which also allowed assign-whilst-paused). And yet, it never felt like it was getting in the way or completely ruining that aspect of the game because it wasn't used for fine-tweaking.

Most of the time when playing in RetroLemmini (which is currently my preferrd engine for actually playing Lemmings), I click to cancel my previous replay because I want to mostly do the solution over and don't care about sacrificing a few assignments; a fresh start is preferable. Again, I cannot be the only player who thinks like this.

Conversely - when I'm playing in NeoLemmix/Lix, I use backwards framestepping, of course - why wouldn't I. But, it makes the game feel totally different. Specifically: there are no real-time stakes at all. Nothing I do really matters because I can just skip back and redo it immediately. It feels like I'm editing a video clip rather than playing a video game. This is fundamentally a different way to play, and it matters.

Is that closer to the holistic argument you were hoping for? I'd be happy to unpack my thoughts on this further. And make no mistake: I would never want the player assists to be removed from Lix, NeoLemmix or even SuperLemmix. I just want to be very careful about adding them to RetroLemmini.

Flopsy has reported that this bug is still present in NLCE.

Help welcome.

Looking at the video again, this is a repeat of this bug in SuperLemmix. I never did figure out exactly what was causing it because it's impossible to test for. It happens randomly, and there are no definite steps that will cause it to happen. I can click many times in the Level Select menu and not duplicate the reported behaviour.

So, unless anyone has any debugging suggestions, this might be another one we just have to live with.

Lix 0.10.33

I played a singleplayer level with save requirement 20/20. I put one blocker, saved the 19 other lix, then nuked the blocker and announced to the livestream viewers that I had won the level.

No! The end-of-level screen should look sufficiently different after a loss than after a win. Maybe winning needs a big green checkmark that flies into view and smacks onto the table of statistics. This is obviously kitsch, but it shows the degree of difference that we might want. Ideally, I'd show the button for the next level on the end-of-level screen, still. Or should there be an extra button first to click that makes the next level appear as a button on the end-of-level screen?

Or: Nuking a lost level shouldn't end the map. But how should people quit the level with the mouse then ...

Winning looks like losing, what an embarrasing UI fail. The most fundamental aftermath of a game (singleplayer or multiplayer) is immediate clarity of whether you've cut the mustard or not.

-- Simon

In Lix, since version 0.10.12 from June 2023, you can choose to always be in insert mode. (By default, you're only in insert mode when the tweaker is open.)

I configured Lix to always-insert. Then I played Flopsy's yet-unreleased levels (easier ones from SEBLems and Gotta Go Fast) for almost 3 hours on livestream. I got confused at least once, IIRC even a second time. Both times, I didn't have the tweaker open. How exactly did I get confused? I'll have to re-watch the 3-hour vod and tell you in detail. This is the playtest that I promised.

For now, I can't yet fully recommend to make it a single mode. I didn't change the Lix option defauls yet, either. In Lix 0.10.34 from yesterday, the default is still that you're in insert mode if and only if the tweaker is open.

I'll still have more to answer from all your posts.

-- Simon

Congratulations!

In DOS Lemmings 1's We All Fall Down, I now believe that digger-digger cancelling isn't possible in a way to produce a permanent staircase. Your 2-digger solution is thus NeoLemmix-only. The best in DOS is to send a few lemmings at once  through a single digger hole in progress. Some DOS releases have the wrong fall distance where the level becomes a 0-skill solve.

Yes, zoom into the relevant details as you explain them.

Scrolling was a tad wild. Consider to scroll ... hm, not necessarily less, but ... more peacefully and deliberately, overall. It can be hard to follow otherwise. Give the audience a second to catch up after scrolling and zooming. Explain during the stable view. Wild pointing with the mouse is perfectly fine during a stable view.

Looking forward to the show!

-- Simon

Lemmings was good in 1991 despite lack of action replay, not because of lack.

I'll rob you for $5000 and give you back $2000 as a reward for not telling the police. That's a good reward, $2000, it's a lot more than $1000. This is why I wrote: instead of punishing/locking. Different feeling.

Lemmings 3 happens to implement what you're considering. Solutions from an action replay don't count ("learner mode") for progression. Feature is meticulously implemented, but you're not supposed to use it until you've proven your worth. Reaction by Kieran:

Quote from: kieranmillar on July 17, 2023, 10:54:20 PMusing the replay feature makes the solve not count (why oh why did they do this? Idiots)

What I really hoped to see was a holistic argument. Why do all those design decisions make the best possible whole? I'm sure that, yes, you can continue to tell me nice stories about every individual choice however it turns out. X is easy to implement, Y easy to understand, Z is already implemented so it would be more work to cut it and some users would act up, W is optional control feature and we'll continue to support both yes W and no W, ..., but why is this exact collection of choices/features the overall best-suited?

I'm not even arguing that you have to implement rewind at all costs. Maybe drudgework is exactly the right cost for what you envision. Unlikely but possible. Unlikely enough to seek alternatives.

-- Simon

Quote from: Simon on August 23, 2026, 07:20:26 PMIf this means backward and forward skips (+1/−1 physics update, staying paused): You'll have to remove some cheating features from RetroLemmini. As-is, you can click to advance by 1 physics update during pause.

Yes, as I said, it's unfortunate that this is already a part of RL. It won't be getting removed, but it has been made optional by another user/developer (Charles) who seemed to have a similar opinion.

Quote from: Simon on August 23, 2026, 07:20:26 PMthe difference between cheating and not cheating ... is drudgework. This is Mindless's "It's just tedious, and that doesn't make for a good game."

Hm. Maybe I disagree that it doesn't make for a good game. If that's true, why are we all here? The vast majority of original Lemmings ports don't even have a replay system, let alone per-frame control over the gameplay. It can't be that bad a game if there's an entire Forum dedicated to it...

Quote from: Simon on August 23, 2026, 07:20:26 PMYou can design the game like this. But is that good?

Maybe this is a question to which I don't think the answer should automatically be "no". If the player is punished for making a mistake, the real-time gameplay stakes are higher. Some players genuinely enjoy this. Lemmings itself was more or less built on this very principle.

And yes, it has evolved - and that's good. But, we have engines with player assists which we all (myself included) enjoy; we should also have one that presents the more traditional gameplay style.

Quote from: Simon on August 23, 2026, 07:20:26 PMDon't punish/lock X, instead reward Y.

Reward pause-free playing.

Yes, this could work. The Talisman system in Neo/Super is good; both players and designers benefit from it. I thought about implementing a challenge/trophy system in RL but realised how complicated it would make everything and so abandoned it (for now). I think it would only be worth the effort if it were a somewhat online feature, where players could see which trophies they (and other players) had achieved.

Counter-point, though: offering 'modern' features after a successful 'classic' playthrough is 'reward' rather than 'punish', is it not? We're saying: "well done, you completed the level without player assists, now they're unlocked for this level". Maybe it is a bad idea, I'm not 100% convinced of it myself. I'm just exploring the possibilities at the moment.

QuoteDirection-select and assign-whilst-paused are arguably QOL features.

Directional select can allow extra things. In a sufficiently stone-aged engine, you can't assign to certain lemmings in a dense pack unless you force direction. It would enable new things in L2.

Quote from: WillLem on August 23, 2026, 06:25:19 PMFramestepping, on the other hand, is tantamount to cheating

If this means backward and forward skips (+1/−1 physics update, staying paused): You'll have to remove some cheating features from RetroLemmini. As-is, you can click to advance by 1 physics update during pause.

If this means rewind only: Then the difference between cheating and not cheating is that the non-cheater must mind-numbingly fast-forward to get where he wants, and the cheater may skip this fast-forwarding and immediately continue where he wants. Both end at the same game situation. The only difference is drudgework. This is Mindless's "It's just tedious, and that doesn't make for a good game."

You can design the game like this. But is that good?

QuoteA player must complete the level at least once in 'Classic Mode' before the modern controls are unlocked for that level.

Don't punish/lock X, instead reward Y.

Reward pause-free playing. Framestepping enters pause and therefore disqualifies you from winning pause-free, you'll only get a normal win. That would directly support what you seem to want, and it would be a fresh take on the whole matter.

-- Simon

It's IN!
Sunday 4.10.26, 11:43 am UK time!