Page 1 of 3

Golly 3.2

Posted: July 5th, 2018, 11:44 pm
by Andrew
Golly 3.2 is now available from sourceforge:

https://sourceforge.net/projects/golly/ ... golly-3.2/

Main changes to 3.2b2:

* 3D.lua is faster, mainly due to a new ovtable command which provides faster table versions
for some overlay commands (thanks to Chris Rowett).

* Updates and corrections to the Life Lexicon (thanks to Dave Greene).

Re: Golly 3.2

Posted: July 6th, 2018, 5:26 am
by muzik
unit-fraction-orthogonal-spaceships still seems to be using an outdated version of the RainbowASO rule, here's the most up-to-date version:
viewtopic.php?f=11&t=2899#p61655

Re: Golly 3.2

Posted: July 9th, 2018, 2:16 pm
by _zM
bgolly seems to ignore the -i option:

Code: Select all

This is bgolly 3.2 Copyright 2005-2018 The Golly Gang.
- ./bgolly -a Generations -m 64 -i 16 -o star3.rle star2.rle
0: 1,252,639
1: 1,252,807
2: 1,252,851
3: 1,252,743
...

Re: Golly 3.2

Posted: July 9th, 2018, 2:53 pm
by dvgrn
muzik wrote:unit-fraction-orthogonal-spaceships still seems to be using an outdated version of the RainbowASO rule, here's the most up-to-date version:
viewtopic.php?f=11&t=2899#p61655
Getting that into Golly 3.2 would have required a specific answer to this question. I don't understand in detail what is different about the new rule, and didn't see a quick way to find out. So someone else will have to write a reasonable set of pattern comments in this case.

-- Also I really prefer the interleaved-hyperbola effect of the current pattern, in any case.

As a first-draft fix, I can check in (for Golly 3.3) a change to the existing pattern comments, to point to the up-to-date version mentioned above.

EDIT: Related question: should something different be done for the format of unit-fraction-orthogonal-spaceships.zip, for the next iOS version?

Re: Golly 3.2

Posted: July 9th, 2018, 3:27 pm
by Apple Bottom
Nice, kudos to everyone involved for your excellent ongoing work!

The latest Lexicon fixes don't seem to have made it into the bundled Lexicon in this release though --- for instance, Sir Robin is still listed as (2,1)c/7 in :spaceship:.

Re: Golly 3.2

Posted: July 9th, 2018, 5:29 pm
by dvgrn
Apple Bottom wrote:Nice, kudos to everyone involved for your excellent ongoing work!

The latest Lexicon fixes don't seem to have made it into the bundled Lexicon in this release though --- for instance, Sir Robin is still listed as (2,1)c/7 in :spaceship:.
Darn manual table patching system anyway. I suspect that it's only the tables that may sometimes be not quite up to date -- and really only the spaceship and Herschel tables have seen many changes, so they're the most likely to have stubborn typos.

Basically, the updates made it in into the HTML, but then got stomped on by the not quite up-to-date contents of manual-edits.txt (in the source distribution). Will fix and check in new HTML files, to be ready for when the next Golly build happens -- and new ZIP files for the regular Lexicon.

Anyway, I'm quite sure that I rebuilt all the HTML files from the latest source, not too many days ago and since any major new definitions or changes went in, and I did enough spot checking to be sure that at least some of the latest stuff was there.

For example, :GPSE: and :BLSE: made it in, and they were in the final couple of weeks' changes. Do you see any other "for instance"s, besides the Sir Robin (2,1)c/7 problem?

Re: Golly 3.2

Posted: July 9th, 2018, 8:05 pm
by Andrew
_zM wrote:bgolly seems to ignore the -i option: ...
Which platform are you using? 32-bit or 64-bit? I can't reproduce the problem on my Mac:

Code: Select all

This is bgolly 3.2 Copyright 2005-2018 The Golly Gang.
- ./bgolly -m 64 -i 16 -o foo.rle Patterns/Life/Methuselahs/rabbits.lif
0: 9
16: 13
32: 28
48: 36
64: 56
EDIT: I *can* reproduce the bug if I use "-a Generations" and run a file in Patterns/Generations. I'll look into it.

Re: Golly 3.2

Posted: July 9th, 2018, 11:35 pm
by Andrew
_zM wrote:bgolly seems to ignore the -i option: ...
Actually not a bug. bgolly ignores -i (and -2) if the given pattern is bounded, which I'm guessing your pattern was. Here's a bgolly run using an unbounded Generations pattern:

Code: Select all

This is bgolly 3.2 Copyright 2005-2018 The Golly Gang.
- ./bgolly -m 64 -i 16 -a Generations Patterns/Generations/Delta.rle
0: 65
16: 96
32: 79
48: 114
64: 213
bgolly should probably print out some sort of warning rather than silently ignore those options.

Re: Golly 3.2

Posted: July 10th, 2018, 3:23 am
by _zM
Andrew wrote:
_zM wrote:bgolly seems to ignore the -i option: ...
Actually not a bug. bgolly ignores -i (and -2) if the given pattern is bounded, which I'm guessing your pattern was.
Yep, that was it. Thanks for letting me know.

Re: Golly 3.2

Posted: July 10th, 2018, 7:44 am
by Apple Bottom
dvgrn wrote:For example, :GPSE: and :BLSE: made it in, and they were in the final couple of weeks' changes. Do you see any other "for instance"s, besides the Sir Robin (2,1)c/7 problem?
Nope --- that's the only one I spotted (and only because I'd fixed that myself in the source txt).

What's the process for building the Lexicon and getting it into Golly, anyway? (Please do feel free to send me a PM rather than replying here, BTW, as it's not really on-topic.)

Re: Golly 3.2

Posted: July 10th, 2018, 8:31 am
by dvgrn
Apple Bottom wrote:What's the process for building the Lexicon and getting it into Golly, anyway? (Please do feel free to send me a PM rather than replying here, BTW, as it's not really on-topic.)
One more message won't hurt, I think. If someday I move back to my cabin in the Adirondacks where there's no Internet available (as I am occasionally tempted to do -- it would make life, if not Life, much simpler!) then there might be some hope that Lexicon releases might still happen... if these processing instructions are reasonably comprehensible.

The process is currently an unholy mishmash of scripts -- a Python script (lexicon-word-wrap.py) to word-wrap the raw text, then Stephen Silver's Bash script (makehtml.sh) calling C code (perfect.c and htmlize.c) to make the HTML files, and finally a Perl script (modify.pl) on a copy of the resulting multi-page HTML files to adjust them into Golly's preferred format.

Then after that there are six tables that none of the scripts handle correctly, because they get treated like patterns. I've just corrected them by hand for the time being, and put the copies in a file called manual-edits.txt, which is in the lex_src.zip archive. That's where the Sir Robin problem crept in -- I didn't remember to update the corrected table to match your (2,1)c/6 fix. There are various other details that the conversion code doesn't handle correctly, which are also listed in manual-edits.txt.

Of course there's a vague plan to convert all of this mess into a single Python or Lua script eventually, but at any given moment it's not worth the effort to do all the porting work. There's bound to be an uncomfortable period with mysterious new bugs. At the moment the great advantage of the current process is that it does actually work... eventually... mostly.

EDIT One obvious improvement will be to include manual-edits.txt at the end of the online raw copy of the Lexicon, to make it easier to remember to fix those corrected tables whenever they change. I'll do that for Release 30 (but probably won't get around to porting any of the other code).

Re: Golly 3.2

Posted: July 10th, 2018, 9:46 am
by dvgrn
A new bug report:

Folks on Discord report that when you left-click and middle-click at the same time in the Windows version of Golly, it either

A) crashes Golly (I haven't reproduced this), or

B) puts Golly in the weird state that I have described before, but didn't know how to reproduce: clicks on menu bars work like clicks on the underlying grid, which usually means that the screen starts scrolling madly away from wherever the mouse is (because the position is at-or-beyond the edge of the viewable area.)

I had gotten as far as figuring out that it seemed to happen when my hand slipped on the mouse a little bit. But for some reason I never managed to find that the problem was a simple simultaneous left- and center-click.

-- Now if steps can be found to reproduce the remaining Undo bug, we'll really be getting somewhere! Undo is definitely less buggy these days, but especially after marathon editing sessions I do occasionally get Golly into a strange state where an Undo operation will change cells that are offset by a small-to-moderate distance from the correct ones. I've seen this in Golly 3.1, but not in 3.2 yet.

Re: Golly 3.2

Posted: July 10th, 2018, 11:18 am
by _zM
This is what happens to Golly for me after a left+middle mouse click (Xubuntu 16.04 LTS). Notably, using the mouse becomes impossible not just in Golly, but also in any other programs that might be running.

Re: Golly 3.2

Posted: July 14th, 2018, 9:16 pm
by Andrew
_zM wrote:This is what happens to Golly for me after a left+middle mouse click ...
This bug has been fixed for the next version (thanks to Chris Rowett).

Re: Golly 3.2

Posted: July 16th, 2018, 12:49 am
by Hooloovoo
As far as I can tell, the version of golly in most linux distributions is ancient (at least gentoo and debian have 2.8, and android has 2.6 iirc). Are the maintainers of the Golly packages generally associated with the community, or the distro? I think it would be worth it to have community members who use a given distribution to step up.

Re: Golly 3.2

Posted: July 16th, 2018, 4:38 am
by wildmyron
Hooloovoo wrote:As far as I can tell, the version of golly in most linux distributions is ancient (at least gentoo and debian have 2.8, and android has 2.6 iirc). Are the maintainers of the Golly packages generally associated with the community, or the distro? I think it would be worth it to have community members who use a given distribution to step up.
Linux packages are generally maintained by people affiliated with the distribution as opposed to the software itself, although there can certainly be overlap between these groups of people. For Android and iOS the Golly Gang publish apps on the respective app stores. Andrew has said that he will be looking at updating the Android app following the release of Golly 3.2.

For the Linux distributions you can find out who maintains the package by looking at the package directory for the distribution, e.g.:

Debian - https://packages.debian.org/sid/golly
Ubuntu (generally follows Debian) - https://packages.ubuntu.com/cosmic/golly
Gentoo - https://packages.gentoo.org/packages/app-misc/golly
Fedora - https://apps.fedoraproject.org/packages/s/golly (and someone filed a bug to update to 3.1, but no action)

Re: Golly 3.2

Posted: July 28th, 2018, 9:02 am
by NickGotts
Is there a separate 64-bit version, as there was for earlier Gollys?

Re: Golly 3.2

Posted: July 28th, 2018, 9:51 am
by wildmyron
The SourceForge paged linked to in the first post of this thread has all the usual versions including 64bit versions for Windows and Linux. The Mac version is a universal binary, I believe.

Re: Golly 3.2

Posted: July 28th, 2018, 12:26 pm
by NickGotts
Thanks, for some reason the other versions didn't show up the first time I went to that page.

Re: Golly 3.2

Posted: August 10th, 2018, 7:07 pm
by muzik
What exactly is planned for Golly 3.3?

Re: Golly 3.2

Posted: August 15th, 2018, 2:08 am
by 77topaz
I noticed something in the documentation for DRH-oscillators.rle:

Code: Select all

#C ----------------------------------------------------------------------
#C 15240.0.0   "p120-based PRNG"  [DRH 3/21/92]  Pseudo random number
#C             generator, based on an earlier pattern by RWG (a larger
#C             version of 40894.0.0).  Produces a sequence of bits b[n],
#C             represented by gliders, satisfying the recurrence
#C             b[n] = b[n-1] XOR b[n-7].  The sequence has period 127,
#C             so the period of the pattern is 120*127.  Increasing the
#C             separation between the NW and SE parts by 15N units
#C             diagonally changes the 7 above to K=7+N  (N >= -4).  The
#C             period of the sequence depends rather erratically on K:
#C 
#C             K | 2 3  4  5  6   7  8  9  10   11   12   13    14    15
#C             P | 3 7 15 21 63 127 63 73 889 1533 3255 7905 11811 32767
#C 
#C             K |  16  17     18     19     20   21      22      23
#C             P | 255 273 253921 413385 761763 5461 4194303 2088705
#C 
#C             K |      24       25     26        27       28        29
#C             P | 2097151 10961685 298935 125829105 17895697 402653181
#C 
#C             K |       30      31   32   33
#C             P | 10845877 2097151 1023 1057
#C 
#C             Also see 40894.0.0.  (In 1996, DRH built a p30-based PRNG.)
#C             This uses a "boat-bit", a reaction in which gliders hitting
#C             a snake alternately create and destroy a boat. 
#C ----------------------------------------------------------------------
That erratic sequence is precisely A046932 in the OEIS.

Re: Golly 3.2

Posted: August 15th, 2018, 7:05 am
by muzik
On the topic, can we have an updated version of DRH_oscillators added?

Re: Golly 3.2

Posted: August 21st, 2018, 12:34 am
by KittyTac
Is there a reason why you can't draw outside of the billion-cell boundary?

Re: Golly 3.2

Posted: August 21st, 2018, 9:18 am
by Redstoneboi
are we able to update golly without having to delete and reinstall?

Re: Golly 3.2

Posted: August 21st, 2018, 9:30 am
by Andrew
KittyTac wrote:Is there a reason why you can't draw outside of the billion-cell boundary?
The internal setcell routine used by Golly accepts 32-bit signed integers for the x,y coordinates. This means we could have allowed drawing within −2,147,483,648 to +2,147,483,647 but it seemed a lot easier to remember (and document) limits of +/- 1 billion. Happy to reconsider this decision if anyone can come up with a good reason for wanting to draw outside those limits.