Golly 4.1b1

For general discussion about Conway's Game of Life.
Post Reply
User avatar
Andrew
Moderator
Posts: 831
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Golly 4.1b1

Post by Andrew » March 30th, 2021, 3:26 am

The 1st beta version of Golly 4.1 is now available for testing:

http://www.trevorrow.com/golly/Golly-4.1b1-Mac.dmg (for macOS 10.10+)
http://www.trevorrow.com/golly/golly-4. ... bit.tar.gz (for 64-bit Linux)
http://www.trevorrow.com/golly/golly-4. ... -64bit.zip (for 64-bit Windows 7+)

Changes to 4.0:

* The Mac app is a universal binary that runs natively on Intel or Silicon. If you have one of the new Macs with an M1 chip then to run Python scripts you'll need to download the universal2 installer from https://www.python.org/downloads/release/python-392/. Please let me know if it works!

* The cell population inside and outside of a selection can now be displayed in the status bar. Tick the new "Show selection population" check box in Prefs > Edit to enable this feature (default is disabled).

* The g.setoption command now recognizes "showcellborders" for showing or hiding cell borders when zoomed in > 2x.

* Fixed a bug in Mac app when drawing while a pattern is generating (only on Retina screens).

* Fixed a bug in Larger than Life weighted neighborhoods with negative counts.

User avatar
rowett
Moderator
Posts: 2392
Joined: January 31st, 2013, 2:34 am
Location: UK
Contact:

Re: Golly 4.1b1

Post by rowett » March 30th, 2021, 5:05 am

If you're having trouble downloading from the links above try these:

https://lazyslug.com/golly/Golly-4.1b1-Mac.dmg (for macOS 10.10+)
https://lazyslug.com/golly/golly-4.1b1-gtk-64bit.tar.gz (for 64-bit Linux)
https://lazyslug.com/golly/golly-4.1b1-win-64bit.zip (for 64-bit Windows 7+)

Naszvadi
Posts: 821
Joined: May 7th, 2016, 8:53 am
Contact:

Re: Golly 4.1b1

Post by Naszvadi » March 31st, 2021, 12:34 pm

rowett wrote:
March 30th, 2021, 5:05 am
If you're having trouble downloading from the links above try these:

https://lazyslug.com/golly/Golly-4.1b1-Mac.dmg (for macOS 10.10+)
https://lazyslug.com/golly/golly-4.1b1-gtk-64bit.tar.gz (for 64-bit Linux)
https://lazyslug.com/golly/golly-4.1b1-win-64bit.zip (for 64-bit Windows 7+)
Thanks! Which commit ID had been tagged az "4.1b1"? And I would like to report a bug, HexGrid.lua vs 4.1b1 floods patterns with rules defined by ruleloader, especially this:

Code: Select all

#C x = 20, y = 15, rule = cgolinhexa:T20,16
x = 20, y = 15, rule = cgolinhexa
BDBDBDBDBDBDBDBDBDBD2$BDBDBDBDBDBDBDBDBDBD2$BDBDBDBCBDBDBDBDBDBD2$BDB
DBDACBDBDBDBDBDBD2$BDBDBDBDACBDBDBDBDBD2$BDBDBDBDBDBDBDBDBDBD2$BDBDBD
BDBDBDBDBDBDBD2$BDBDBDBDBDBDBDBDBDBD$!
Host OS is linux, custom built golly commit ID and Lua version is:

Code: Select all

user@errorlevel:0:~/git.code.sf.net/p/golly/code$ git tree -1
* 1fdac2b (HEAD -> master, origin/master, origin/HEAD) Fixed a bug in Mac app when drawing while a pattern is generating (only on Retina screens).
user@errorlevel:0:~/git.code.sf.net/p/golly/code$ lua -v
Lua 5.2.4  Copyright (C) 1994-2015 Lua.org, PUC-Rio
user@errorlevel:0:~/git.code.sf.net/p/golly/code$ 
Golly reported version:

Code: Select all

This is Golly version 4.1b1

© 2005-2020 The Golly Gang:
Tom Rokicki, Andrew Trevorrow, Tim Hutton, Dave Greene, Chris Rowett,
Jason Summers, Maks Verver, Robert Munafo, Dongook Lee.
In memory of John Horton Conway, 1937-2020.

Golly is an open source, cross-platform application for
exploring the Game of Life and other cellular automata.

http://golly.sourceforge.net/

User avatar
Andrew
Moderator
Posts: 831
Joined: June 2nd, 2009, 2:08 am
Location: Melbourne, Australia
Contact:

Re: Golly 4.1b1

Post by Andrew » March 31st, 2021, 11:09 pm

Naszvadi wrote:
March 31st, 2021, 12:34 pm
Which commit ID had been tagged az "4.1b1"?
Sorry, I don't bother tagging any commits.
And I would like to report a bug, HexGrid.lua vs 4.1b1 floods patterns with rules defined by ruleloader ...
Thanks for the report. Bug fixed in this version:
https://sourceforge.net/p/golly/code/ci ... format=raw

User avatar
GUYTU6J
Posts: 1409
Joined: August 5th, 2016, 10:27 am
Location: 拆哪!I repeat, CHINA! (a.k.a. 种花家)
Contact:

Re: Golly 4.1b1

Post by GUYTU6J » May 24th, 2021, 8:50 am

The following problem I found one half year ago still exists now with Golly-4.1b1-win-64bit and Python 3.9.5:
GUYTU6J wrote:
November 7th, 2020, 5:46 am
Curious behaviour related to (once again) synthesise_patt_v1.2.py: Golly 4.0 with python 3.9 seems to look up the list of 464,745 collisions twice, yet still finding the same number of results [compared with Golly 3.4 & python 2.7.16, 64-bit, I think] for a given pattern after 929,490 collisions tried.
Can anyone reproduce the behaviour with the files here (there are still 464,745 collisions, right?) and find the cause? Even if the speed is fast, seemingly repeated scanning could slow down one's pace when doing glider syntheses.
Lifequote:
In the drama The Peony Pavilion, Tang Xianzu wrote: 原来姹紫嫣红开遍,似这般都付与断井颓垣。
(Here multiflorate splendour blooms forlorn
Midst broken fountains, mouldering walls.)
I'm afraid there's arrival but no departure.
Chinese people, enjoy your 🥩🥚🥛 and 🐟!

User avatar
dvgrn
Moderator
Posts: 8012
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Golly 4.1b1

Post by dvgrn » May 24th, 2021, 2:46 pm

GUYTU6J wrote:
May 24th, 2021, 8:50 am
Can anyone reproduce the behaviour with the files here (there are still 464,745 collisions, right?) and find the cause? Even if the speed is fast, seemingly repeated scanning could slow down one's pace when doing glider syntheses.
That link is to the old repository, which I'm planning to get rid of at some point. Please use the version at this link instead.

(I'm pretty sure the files were exactly the same until this morning, but I'd rather the "b3s23life_old" link didn't get spread around any more -- so once this all gets sorted out, I'll probably edit your post to remove the old link.)

I've gone ahead and updated the new three-glider-collisions/synthesise-patt to work better with Python3.

I also put in an ugly patch for the doubled counts that were getting reported by the previous version. It wasn't really reading the file twice; there's just some awkward hiccup in the data file, and/or maybe the encoding/decoding process, such that each line gets read as a regular line of data, plus an empty line containing just an ASCII character 10 (which of course never matches the selected pattern).

User avatar
GUYTU6J
Posts: 1409
Joined: August 5th, 2016, 10:27 am
Location: 拆哪!I repeat, CHINA! (a.k.a. 种花家)
Contact:

Re: Golly 4.1b1

Post by GUYTU6J » May 24th, 2021, 10:12 pm

dvgrn wrote:
May 24th, 2021, 2:46 pm
...
I also put in an ugly patch for the doubled counts that were getting reported by the previous version. It wasn't really reading the file twice; there's just some awkward hiccup in the data file, and/or maybe the encoding/decoding process, such that each line gets read as a regular line of data, plus an empty line containing just an ASCII character 10 (which of course never matches the selected pattern).
That said, can I understand that Golly 4.0+ uses a different file-reading mechanism than older Golly, where empty lines that previously ignored will be scanned?
Lifequote:
In the drama The Peony Pavilion, Tang Xianzu wrote: 原来姹紫嫣红开遍,似这般都付与断井颓垣。
(Here multiflorate splendour blooms forlorn
Midst broken fountains, mouldering walls.)
I'm afraid there's arrival but no departure.
Chinese people, enjoy your 🥩🥚🥛 and 🐟!

Null
Posts: 24
Joined: April 21st, 2021, 6:01 am

Re: Golly 4.1b1

Post by Null » May 25th, 2021, 1:36 am

GUYTU6J wrote:
May 24th, 2021, 10:12 pm
dvgrn wrote:
May 24th, 2021, 2:46 pm
...
I also put in an ugly patch for the doubled counts that were getting reported by the previous version. It wasn't really reading the file twice; there's just some awkward hiccup in the data file, and/or maybe the encoding/decoding process, such that each line gets read as a regular line of data, plus an empty line containing just an ASCII character 10 (which of course never matches the selected pattern).
That said, can I understand that Golly 4.0+ uses a different file-reading mechanism than older Golly, where empty lines that previously ignored will be scanned?
No, it's a Python2/3 difference, not a Golly difference.
Edit: I was in haste, but at least you can see the big picture (I may have small details wrong)
Edit: Here's the relevant SO link: https://stackoverflow.com/questions/4599936

User avatar
dvgrn
Moderator
Posts: 8012
Joined: May 17th, 2009, 11:00 pm
Location: Madison, WI
Contact:

Re: Golly 4.1b1

Post by dvgrn » May 25th, 2021, 8:58 am

Null wrote:
May 25th, 2021, 1:36 am
Edit: Here's the relevant SO link: https://stackoverflow.com/questions/4599936
Thanks -- that allows for a much less kludgey fix for the Python3 processing code. Somehow both of the data files ended up with completely non-standard \r\r\n newlines, and Python was somehow translating that into twice as many lines of data as actually existed, with alternate lines being just a single \n character. Not quite sure how it did that, since the order isn't right if it's interpreting the file in old-Mac-newlines mode.

The Wikipedia entry on "Newline" is very entertaining, by the way -- there are quite a few newline markers catalogued there that were new to me... but it seems nobody has ever used \r\r\n before!

Anyway, the code below is a decent solution for now, without rebuilding the BZIPped data files -- I've checked this in as synthesise_patt_v1.4.py:

Code: Select all

patts = open("rles.txt","r", newline="\r\n")
cols = open("colseqs.txt","r", newline="\r\n")
Really the next step for synthesise-patt is to rebuild the entire database using octohashes instead of population sequences. That will get rid of all (or at least almost all) of the false-positive matches that currently come up when the pattern you're looking for is just a tick or two from stabilization.

While we're at it we should extend the database to the 4-glider collisions that wildmyron added for the synthesise-collision search. If anyone wants to tackle that problem, please start a new thread and I'll help out with any tricky spots -- and also transfer any remaining octohash processing code that I haven't managed to publish yet.

Post Reply