Inflection point for updates to gun collection

[dvgrn]

There's been enough discussion on the glider-gun collection thread now
( https://www.conwaylife.com/forums/viewtopic.php?f=2&t=1657 )
that it's getting a little harder to find the point where the GitHub repo stopped being updated to match new discoveries posted to the thread.

It looks like Sokwe's 24 April 2019 gun improvements are in the collection --

https://www.conwaylife.com/forums/viewtopic.php?p=75422#p75422

-- but the following post by Kazyan on the same day contains guns not in the collection:

https://www.conwaylife.com/forums/viewtopic.php?p=75423#p75423

I haven't double-checked Kazyan's follow-up 42-gun post and other gun improvements on the next page, but it looks like those also aren't in the collection.

[ceebo]

Hello! I've got so far out of touch CGOL that I stopped looking at the forum entirely. There was some unbeliveable stuff with glider synthesis last time I looked. Also with the reverse caber tosser... but I never had time to figure out how that worked exactly.

As for the gun collection... yes I agree it's hard to work out the point where the updates stopped. When I get time I'll look at your links above and see if they bring back any memories.

[dvgrn]

I could try to do some recruiting for someone, or maybe a small team, to put together the pull requests needed to get the gun collection back up to date. Looks like Ian07 has done a trial one, but probably no one will do any more until that one is incorporated successfully.

The black-magic part of this project still makes me pretty nervous. If I ever knew exactly what all the options were for parameters like tentative_compression, y_slack, y_trips, x_overhang, weird_y, or osc_mod, I've long since forgotten. But I think these days most updates could be successfully done with nothing in the gun RLE but a #compression comment.

Would you suggest I just try starting from the current repo, adding let's say Kazyan's 42 guns (or templates for them) and then running get_area.py -- and just see what happens? Or is there some easily transferrable knowledge that I obviously don't have yet, if I'm saying this?

[ceebo]

I'd say the most important things to know about the python script are:

base period, oscillator period and multipler are embedded in the filename. e.g. p00399osc7x3.rle

compression is the smallest period where the gun is guaranteed to work. If there are no glider crossings then this will often be 78 for a syringe based gun. If there are glider crossings then compression could be much much higher.

tentative_compression is the compression of the gun without considering glider crossings. If the script thinks there could be a record setting gun at a period less than the compression but at least as big as the tentative compression then the gun will be noted in the Tentative Guns section at the end of the README.

I'd say the only other important pieces of black magic are x/y_trips and x/y_slack.

A variable gun with just a single slide will have x_trips = y_trips = 1. In fact x_trips = 1 and y_trips = 1 are the defaults and never need to be specified.

If we have a gun in a V-shape like so:

\     /
 \   /
  \ /

Then it will have x_trips = 1 but y_trips = 2, because the glider bounces up and down twice.

x_slack and y_slack define how many extensions of gun are for "free" in a particular direction.

Take the example of the V-shaped gun above and imagine the right hand arm finishes 5 cells below the left hand arm. Then y_slack would be equal to 5 because the first 5 extensions cause no increase in the y-direction.

Say the original dimensions were W x H then with y_trips = 2 and y_slack = 5 you would get the following dimesions for the first 9 extensions:

Extension	x	y
0	W	H
1	W+1	H
2	W+2	H
3	W+3	H
4	W+4	H
5	W+5	H
6	W+6	H+1
7	W+7	H+1
8	W+8	H+2
9	W+9	H+2

The stuff such as overhang, weird and osc_mod can probably be forgotten about.

I know this still doesn't answer your question about the inflection point.