Download the VODs as you like.

I'll have to make up my mind about rehosting. In all strictness, I play background music from random sources without permission. Unsure yet if I want that on Youtube.

-- Simon

Quote from: Simon on June 10, 2026, 05:00:24 PMRecording will remain for 14 days at: https://www.twitch.tv/simonnaar

Do I have your consent to download your VODs and upload them somewhere so people can watch them after the 14 days are up?

Thanks! I promise I'll re-test performance on Turrican's huge map 10,000 AD from Level of the Year 2024 with disabled decorations. Prod me after release if I forget.

-- Simon

I did enjoy the pack, thanks! I'd agree with the designation of 'Easy' but there are still a few tricky levels in here (18, 50, 55, 62, 69, 83, 90 and 94 in particular, for me). I found the trickiness was usually down to the awkwardness of the terrain and needing to get to all different parts of the screen with the skills available.

I do like this kind of pack as well as hard ones because you can just solve a little puzzle or two, go away and do something else and come back to it. Bite size entertainment! With some packs, you can end up spending well over half an hour on a single level, which has its own draw, but sometimes this is the way to go! Also liked your use of varied tilesets throughout

Replays attached. I dunno if you're someone who is keen to enforce intended solutions, but there're probably quite a few backroutes in here.

Wed 10 stream is over!

Recording will remain for 14 days at: https://www.twitch.tv/simonnaar

-- Simon

I did not encounter any problems with level 68. The Crystal Style has two (2) options for exit. See screenshot attached. I have chosen the other one (red vortex). Both are working on my pc.
Thanks for the warm welcoming message .
Enjoy the pack.

Re: the C# used in NLEditor for the Random ID generator

Quote from: WillLem on June 10, 2026, 03:31:10 PMThe bit you're interested in is this:

Code Select Expand

        private void btnRandomID_Click(object sender, EventArgs e)
        {
            CurLevel.LevelID = (ulong)Utility.Random().Next() +
                              ((ulong)Utility.Random().Next() << 32);
            txtLevelID.Text = CurLevel.LevelID.ToString("X16");
        }


Yes, interesting! I'm not familiar with the Random() method in C# but I'm guessing it must do automatic seeding. In C++, if you use srand() from the standard library, you need to manually seed it with a parameter otherwise it would generate the same random number every time the program is run. I assume '<<' in this context is a left-hand bitwise shift in the same way as in C++.

Quote from: JawaJuice on June 10, 2026, 03:49:17 PMThe only time I ever used Delphi at work was back in the early Noughts ... I've sure the language has evolved out of sight since I last looked at it!

NeoLemmix and SuperLemmix are both Delphi-based, which is pretty much the only reason I know it exists. I find its form-building interface very comprehensive and user-friendly, and it's still a great IDE for simple tool building.

Thanks for reporting, added this to the list for 1.1.2 (hopefully due in the next few weeks).

Quote from: Simon on June 09, 2026, 07:08:16 AMCan you find something better for replay-to-level association that doesn't give such opportunity for mistakes?

Lemmini's replay-to-level matchups have always been led by the level's title and pack position. It has caused problems, particularly with external levels.

RetroLemmini patches the replay matchups for external levels, but still relies on level name and position. Here's the current match flow:

1) Check the DMA levels for a matching name and position -> fallback to external levels if no match is found
2) Check all levels for a matching name and position -> fallback to external levels if no match is found
3) Check all levels (including external) for a matching name only

This is somewhat tighter than (Super)Lemmini(Too)'s replay matching and rarely results in mismatched or unmatched replays.

I did consider implementing an ID system, but given the size of the back catalog it does seem a bit too late (and complex) to introduce this now.

Matching by name is now about 99% successful for individual replays, but yeah for a mass checker it probably does need something a bit more foolproof. Suggestions welcome.