Not that I'm particularly pushing for the skill, but I feel compelled to clarify the confusion you seem to have over glue pourer, at least relative to DOS L2 behavior I've just re-tested.
Glue-pourer: While very powerful, the exact mechanics are not easy to understand, especially when it comes to slopes: On a 30° slope the glue runs down without sticking anywhere. On a 45° slope it runs down for about 8 pixels and then sticks. I still don't know why exactly it does that? If it runs down big steps (like with vertical walls of 8 pixels or so) it never glues, as far as I can tell. At ledges the behavior of a single pixel of glue depends on whether the pixel below it is air, usual terrain or another glue-pixel that just got solid. Not to mention that when running along the glue is twice as fast as a lemming, but somehow slows down when sticking.
Your description seems to mention a lot of behavior that I just don't see when testing in DOS Lemmings 2
:
On a 45° slope it runs down for about 8 pixels and then sticks.
Um, no? The glue solidifies as soon as it hits the slope; it doesn't flow down for a bit and then sudden stops. You do remember that the vertical stream of glue falling down from the bucket starts a few pixels away from where the pourer is standing, right? The solidification spreads to the right starting from exactly where the stream landed, which will be a few pixels away from where the lemming stands. It never flowed from the ground where the lemming is standing into the spot that it is sticking.
If it runs down big steps (like with vertical walls of 8 pixels or so) it never glues, as far as I can tell
That's not exactly true either. The key point is that when the pixel of glue lands and there is still a cliff to the left of it (for a right pourer), whether the step landed on is only 1 pixel wide horizontally versus more than 1. It glues (ie. solidifies into terrain) for the former, but continue flowing right for the latter.
The behavior I just described is ultimately key for the glue's signature behavior, where it creates horizontal glue platform the same vertical thickness as the ledge it is falling off from. That very same behavior also leads to the perhaps slightly surprising side effect of the glue solidifying while moving down slopes 45 degrees or steeper. Removing the behavior would cause the glue to basically always solidify to 1-pixel thick horizontal platforms.
when running along the glue is twice as fast as a lemming, but somehow slows down when sticking
Not sure how you are seeing any kind of slowdown. Each glue pixel either moves 2x speed (or perhaps more accurately, glue movement is processed twice per physics update), or it stops and solidifies. At no point does a glue pixel ever move at 1x speed.
One thing that does happen is that within a single physics update cycle, typically you'll wind up with a whole mass of glue pixels all solidifying into terrain all at once. For example, when flowing down the ledge, none of the glue pixels are terrain until the leading glue pixel finally reaches the bottom of the ledge, and then all at once that whole vertical column of glue becomes solid terrain, causing the rest of the glue to starting falling one further to the right (ie. the new ledge resulted from the glue pixels that just became terrain). Similarly, when glue is flowing flat horizontally towards a wall, none of the glue is terrain until the leading glue pixel hits the wall, then the entire layer of glue starting from where the glue stream landed to the leading glue all at once turn into terrain.
===================
So okay, I'll agree it's somewhat non-trivial, but I think the underlying rules are not that bad for the L2 glue pourer. Here's my stab at it for a right-facing glue pourer (adjust accordingly for left), which I believe creates at least most of the common behaviors people would see:
glue pixel has a current direction of either DOWN or RIGHT.
when updating the glue pixel for movement, turn the glue to terrain in the following 4 cases (T = terrain, . = air, v = glue current direction DOWN, > = glue current direction RIGHT)
Tv
T.
Tv
..
vT
T
>T
T
In all other cases, move glue down one if pixel below is air, else move right one. Update the current direction based on whether you moved down or moved right.
(I strongly suspect you might not even need the concept of current direction at all, but I haven't tested in enough details to confirm.)