golly-test Mailing List for Golly

2009/5/31 Andrew Trevorrow <an...@tr...>:
>> I like how opening a zip from the file menu brings you to the same UI
>> place, this feels right. (Though maybe the back button should be
>> disabled at that point?)
>
> Why should that button be disabled?  I find it very handy to be
> able to go back to a previous help page.

Not permanently disabled, I just meant temporarily grayed out because
there isn't any history at that point, like when you open a new
browser window.


-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/

Tim:

> I've filled out the golly access page for the rule table repository,
> and all the zips are working perfectly without any changes.

Excellent.  It looks like you are only listing stuff that isn't
already included in Golly?  If so, good idea.

Perhaps include a link to the repository website?  I know we have
one in Help > References but a bit of redundancy never hurts.

> (Well, with the exception of simulators.zip which has Clouds.table but
> rule:clouds which doesn't work because of the case sensitivity.)

Mac and Windows users don't see any problem, but for Linux users
someone had better repackage that zip so the correct case is used.
(And I think it's probably nicer to stick with capitalized words
for rule names.)

> The only improvements I can think of:
> 
> 1. The behaviour is different depending on whether the contents of a
> zip or an archive are being shown. If it's an archive, a .table file
> link opens the table straight away, no way to view it in a text
> editor. If it's in a zip, clicking on a .table file opens it in a text
> editor, now way to load it. Could we make this consistent: left-click
> loads it, right-click views it (as in the normal patterns and scripts
> pane). Likewise for .rle files.

Good point.  I'll make those changes.

> 2. Mouse-over a link could show its target URL in the status bar, like
> firefox does (it feels weird clicking into the unknown)

Excellent idea.  Will do.

> I like how opening a zip from the file menu brings you to the same UI
> place, this feels right. (Though maybe the back button should be
> disabled at that point?)

Why should that button be disabled?  I find it very handy to be
able to go back to a previous help page.

Andrew

Tom:

> Are we going to use a different extension for Golly archives packaged as zips?
> 
> Clearly we don't want to usurp ".zip".

Um, why not?  These are normal zip files so I don't see any need
to give them a non-standard extension.  Surely that's just likely
to cause confusion -- plus it makes it more awkward for people to
create such archives.  Or am I missing something?

Perhaps if Golly ever has the ability to *create* special-purpose
zip files then I can sort of see an argument for using a
Golly-specific extension, but until then I don't see any point.

Andrew

Dave:

> Not sure the "autorun" model is quite right -- scripts often have
> names that one might want to keep, and Golly doesn't allow calling one
> script from another (does it?)

I've never tried it but you might be able to do this (from Python)
using the execfile() command.  There is also a system() command in
the os module that might do the job.

> -- Sometimes the point of clicking on a ZIP archive might be just to
> download a local copy of the contents -- not necessarily to do
> anything with them immediately.  Popups asking about running a script
> are an annoyance if you just want to grab copies of utility scripts to
> use later.

Well, in that case just use your browser and download the zip file
in the usual manner.  Golly won't try to run a script if the zip
contains more than one script (or more than one pattern).

> -- We *already* have a very efficient way of exploring a pattern
> archive like 'jslife':  decompress the archive into a subfolder in
> Patterns, then use the treeview as usual.

Yep, that is certainly a sensible approach.  But given that we want
to use zip files as a container format for the simple case of a
pattern+script it seems reasonable to try to handle more complex
zip files in an intuitive way.

> Seems as if that should be
> an option -- i.e., convert the online archive into a local archive.

Yep, I guess we could have that as an option.  I've never tried it,
but apparently it is possible to include a button in a wxHTML window
and link that button to some arbitrary action.  So we could have
a button called "Create Local Archive" that would unzip the entire
archive as a new subfolder in a prompted location.  Might leave this
for a future version (or feel free to investigate it yourself!).

> -- Some scripts have associated patterns, but opening the pattern and
> then running the script doesn't really make sense:  a script similar
> to metafier.py might need a "tile" pattern associated with it, but you
> don't ever want to open the tile pattern by itself.  Also it would be
> best to decompress both files into the same directory, rather than
> sorting them out into Scripts and Patterns subfolders or accessing
> them directly from the archive -- otherwise the script won't be able
> to find the pattern.

Good point.  I'm not really sure what to do in this situation.
I'm sure there will be other cases where the way Golly currently
processes a zip file is not quite flexible enough.  That was one
reason I added the ability to open a clicked text file in the
user's editor.  Golly will do this for any file with a .txt or .doc
extension, or any file containing "readme" in its name (any case).
It will also display any clicked .html file in the help window.
So there are a number of ways to include documentation in a zip
file if you wish to explain how it should be used.

I can think of various ways we could include info within a zip file
to tell Golly how to process the contents, but I'd rather start
simple and go more complicated only if there's a clear need.

Andrew

Thanks for the feedback on zip file handling.  I'll respond to all
your messages later (it's been a hectic social weekend here) but
for now I just want to mention a Windows-only problem I discovered
late yesterday.  Some background first...

As part of the zip file changes, I made a slight change to where
temporary files are created.  On Windows, the folder for storing
user-specific temp files is:

C:\Documents and Settings\your-user-name\Local Settings\Temp

Instead of writing all of Golly's temp files into this folder I now
create a subfolder with the name "gol..." (unique to each instance)
and write all temp files (eg. patterns extracted from zip files) into
that subfolder.  This makes it safer to run multiple instances of
Golly, and allows a zip-extracted pattern file to have the same name
as its zip entry rather than some gobbledygook name which would
look very ugly when shown in the main window's title bar.
When Golly quits, it deletes all files in the subfolder and then
deletes the (empty) subfolder.  This all worked fine on Mac and
Linux, but on Windows the subfolder wasn't being deleted.

I've fixed this problem in my latest commit, but if you've been
using Golly a lot over the last couple of days then you might want
to go into your Temp folder and delete all the gol* subfolders
(make sure you quit Golly first).  While there you might also
see some gol* files left over as a result of Golly crashing,
so remove all those files as well.

Andrew

2009/5/28 Andrew Trevorrow <an...@tr...>:
> Tim:
>
>> ... Could we have that the script runs
>> automatically (if desired) only if it is called 'autorun.py/.pl' (or
>> something) and lies in the root folder of the zip? That would make it
>> clearer, and avoid having scripts run that weren't intended to be.
>
> Yep, we could do something like that, although I'd rather start simple
> and, after testing, see if there's a need for this sort of thing.
> And just to be clear, I assume autorun.p* would only run depending on your
> script security setting -- ie., if you specify "never run a downloaded
> or zip-included script" then it would not run.

Having seen the zip + Online Archives in action I think it's brilliant
as is - scripts ask permission to run, that's enough.

I've filled out the golly access page for the rule table repository,
and all the zips are working perfectly without any changes. (Well,
with the exception of simulators.zip which has Clouds.table but
rule:clouds which doesn't work because of the case sensitivity.)

>
>> And mentioning folders within zips reminds me that that's another
>> thing to think about - the UI would need to allow browsing through the
>> different folders, ideally.
>
> What I've done is show the entire zip file contents using indentation
> to indicate any folder structure.  For example, after downloading
> LifeHistoryRules.zip from the repository the help window shows this:
>
> ----------------------------------------------
> Zip file: /full/path/to/LifeHistoryRules.zip
>
> Contents:
> HistoricalLife.colors    [installed]
> HistoricalLife.table    [installed]
> LifeHistory.colors    [installed]    (overrides file in Rules folder)
> LifeHistory.table    [installed]    (overrides file in Rules folder)
> LifeHistoryClear.colors    [installed]
> LifeHistoryClear.table    [installed]
> ReadMe.txt
> Scripts
>    Perl
>        ToHistoricalLife.pl
>        ToLife.pl
>        ToLifeHistory.pl
>        ToLifeHistoryMarked.pl
>    Python
>        ToHistoricalLife.py
>        ToLife.py
>        ToLifeHistory.py
>        ToLifeHistoryMarked.py
>
> Files marked as "[installed]" have been installed into your rules folder
> (/HD/Golly/golly/src/Test-rules/).
> ----------------------------------------------
>
> All the files are special links and Golly will do the expected thing
> when they are clicked: load a pattern file, run a script, open a .txt
> file in your editor, etc.
>
> I've mostly got zip file handling working as I want -- just a few loose
> ends to tie up.  I hope to commit the changes before the weekend.

>From what I've seen so far you've done a great job! This is a really
useful feature.

The only improvements I can think of:

1. The behaviour is different depending on whether the contents of a
zip or an archive are being shown. If it's an archive, a .table file
link opens the table straight away, no way to view it in a text
editor. If it's in a zip, clicking on a .table file opens it in a text
editor, now way to load it. Could we make this consistent: left-click
loads it, right-click views it (as in the normal patterns and scripts
pane). Likewise for .rle files.

2. Mouse-over a link could show its target URL in the status bar, like
firefox does (it feels weird clicking into the unknown)

I like how opening a zip from the file menu brings you to the same UI
place, this feels right. (Though maybe the back button should be
disabled at that point?)

-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/

Making a polynormal (apparently, although I don't know the degree) curve.

http://img505.imageshack.us/img505/3692/curve.png


I'm wondering if I should remove the ability for the normal WireWorld logic
components to work and make a challenge to make an adder or something.


2009/5/29 Alan Tennant <ala...@gm...>

> A forth state on the original WireWorld to allow tiny switches. Rule and
> example program attached.
>

Are we going to use a different extension for Golly archives packaged as zips?

Clearly we don't want to usurp ".zip".

I wonder if ".gar" is used anywhere (Golly archive).  Maybe there's a
better name?

-tom

>> ... Could we have that the script runs
>> automatically (if desired) only if it is called 'autorun.py/.pl' (or
>> something) and lies in the root folder of the zip? That would make it
>> clearer, and avoid having scripts run that weren't intended to be.
>
> Yep, we could do something like that, although I'd rather start simple
> and, after testing, see if there's a need for this sort of thing.

Not sure the "autorun" model is quite right -- scripts often have
names that one might want to keep, and Golly doesn't allow calling one
script from another (does it?)  Maybe it would work to have an
autorun.txt at the root of the zip archive that tells Golly which of
the scripts or patterns is The One -- but even that may be an
unnecessary complication.

I did have a couple of random thoughts about ZIP archives:

-- Sometimes the point of clicking on a ZIP archive might be just to
download a local copy of the contents -- not necessarily to do
anything with them immediately.  Popups asking about running a script
are an annoyance if you just want to grab copies of utility scripts to
use later.

-- We *already* have a very efficient way of exploring a pattern
archive like 'jslife':  decompress the archive into a subfolder in
Patterns, then use the treeview as usual.  Seems as if that should be
an option -- i.e., convert the online archive into a local archive.

Maybe convert the URL into a Patterns subfolder name, so that any
updates to the archive can automatically go to the same place?  Or
display ZIP files as read-only folders in the treeview, without
actually decompressing it?  (Probably not worth the trouble unless it
happens to be easy.)

-- Some scripts have associated patterns, but opening the pattern and
then running the script doesn't really make sense:  a script similar
to metafier.py might need a "tile" pattern associated with it, but you
don't ever want to open the tile pattern by itself.  Also it would be
best to decompress both files into the same directory, rather than
sorting them out into Scripts and Patterns subfolders or accessing
them directly from the archive -- otherwise the script won't be able
to find the pattern.

Keep the cheer,


Dave

Tony, do you work as a Complex Systems Analyst or is it just a title?


2009/5/27 Tony Smith <ts...@me...>

> Alan, that rule does show some key characteristics that have kept me
> motivated by Generations 345/3/6 for the past seven months.
>
> It even backs up my current interpretation about the synergistic role of
> emergent order in accelerating the spread of deterministic chaos.
>
> Your rule is clearly richer in puffers, richer in spaceship forms and has a
> 3 cell wick stretcher seed which allows a space filler to emerge naturally.
>
> But because there cannot be large patterns with stable edges, it is
> unlikely to have mechanisms for seeding new active centres at a great
> distance.
>
> However it does have some intriguing forms, especially the slow orthogonal
> glider which can form into rakes with a 2 in 7 slope.
>
> Another test it passes for me is that a glancing head on collision between
> the commonest spaceships produces chaotic growth which by iteration
> 10,000 exhibits almost all the patterns that I noticed while running your
> hand crafted demo to 15,000.
>
> x = 8, y = 11, rule = 124567/34/6
> .2E$.2C$D2AD$B2AB$4A3$4.4A$4.B2AB$4.D2AD$5.2C!
>
> The population at 10,000 is 11,238,510 compared to 89,976 / 371,576 for the
> fastest growing seeds I have found / created under Generations 345/3/6 but
> it appears to iterate comparably fast, presumably because hashing works
> efficiently on the rapidly stabilised chaotic mazes.
>
>
> On 27/05/2009, at 5:41 AM, Alan Tennant wrote:
>
> Remove the 2x1 block on the bottom left to get it rolling, the other things
> are just other things that also happen to work in the rule. I know there is
> someone here who is really into this sort of thing and had a really
> inefficient version of this rule.
> <Alan124567s34s6.rle>
>
>
>
>
> Tony Smith
>
> Complex Systems Analyst
>
> Meme Media
>
> Melbourne, Australia
>
> http://www.meme.com.au/
>
>
>
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
> is a gathering of tech-side developers & brand creativity professionals.
> Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, &
> iPhoneDevCamp as they present alongside digital heavyweights like Barbarian
> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
> _______________________________________________
> Golly-test mailing list
> Gol...@li...
> https://lists.sourceforge.net/lists/listinfo/golly-test
>
>

Tim:

> ... Could we have that the script runs
> automatically (if desired) only if it is called 'autorun.py/.pl' (or
> something) and lies in the root folder of the zip? That would make it
> clearer, and avoid having scripts run that weren't intended to be.

Yep, we could do something like that, although I'd rather start simple
and, after testing, see if there's a need for this sort of thing.
And just to be clear, I assume autorun.p* would only run depending on your
script security setting -- ie., if you specify "never run a downloaded
or zip-included script" then it would not run.

> And mentioning folders within zips reminds me that that's another
> thing to think about - the UI would need to allow browsing through the
> different folders, ideally.

What I've done is show the entire zip file contents using indentation
to indicate any folder structure.  For example, after downloading
LifeHistoryRules.zip from the repository the help window shows this:

----------------------------------------------
Zip file: /full/path/to/LifeHistoryRules.zip

Contents:
HistoricalLife.colors    [installed]
HistoricalLife.table    [installed]
LifeHistory.colors    [installed]    (overrides file in Rules folder)
LifeHistory.table    [installed]    (overrides file in Rules folder)
LifeHistoryClear.colors    [installed]
LifeHistoryClear.table    [installed]
ReadMe.txt
Scripts
    Perl
        ToHistoricalLife.pl
        ToLife.pl
        ToLifeHistory.pl
        ToLifeHistoryMarked.pl
    Python
        ToHistoricalLife.py
        ToLife.py
        ToLifeHistory.py
        ToLifeHistoryMarked.py

Files marked as "[installed]" have been installed into your rules folder
(/HD/Golly/golly/src/Test-rules/).
----------------------------------------------

All the files are special links and Golly will do the expected thing
when they are clicked: load a pattern file, run a script, open a .txt
file in your editor, etc.

I've mostly got zip file handling working as I want -- just a few loose
ends to tie up.  I hope to commit the changes before the weekend.

Andrew

> I had no idea what you meant until I googled for "falling game of
> sand" and found some videos, and a song. What fun.
>
> It's the same idea, just a much cruder version. No explosions, sadly.


Explosions and animated characters added in some versions but the
interaction between the sand, earth, salt, oil, water etc... is incredible
and often subtle, the way it slides, bunches is a bit springy etc...


You can download the files immediately, if you want to play:
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/Sand-Margolus-emulated.table
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/Sand-Margolus-emulated.colors
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/Sand.rle


Nice, although I don't like the way it's unpleasent to work with like the
lettuce gas. It's like the gems and rocks game.


Likewise DLA is here:
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/DLA-Margolus-emulated.table
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/DLA-Margolus-emulated.colors
>
> http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/DLA.rle
>

Verry nice, what happens when you make it less selective about when a
particle joins or make the particles bounce at 90 degrees not just back the
excact way they came.

On Wed, May 27, 2009 at 2:08 PM, Alan Tennant <ala...@gm...> wrote:
> I don't plan to upgrade my Golly until the next public release. Can you now
> do "falling game of sand" type things (maybe at a zoom further away that
> 1:1). That game and it variations are amazing.

I had no idea what you meant until I googled for "falling game of
sand" and found some videos, and a song. What fun.

It's the same idea, just a much cruder version. No explosions, sadly.

You can download the files immediately, if you want to play:
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/Sand-Margolus-emulated.table
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/Sand-Margolus-emulated.colors
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/Sand.rle

Likewise DLA is here:
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/DLA-Margolus-emulated.table
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Rules/DLA-Margolus-emulated.colors
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/DLA.rle

Be sure to read the comments under the 'i' button. In particular you
can't just draw on the grid without breaking the emulation pattern -
you need the updated import and export scripts:
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/import.py
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/export.py

If you want to experiment with your own Margolus rule tables you'll
need this script:
http://golly.cvs.sourceforge.net/viewvc/*checkout*/golly/golly/src/Patterns/Other-Rules/Margolus/emulate-Margolus-table.py

Tim

>
> 2009/5/27 Tim Hutton <tim...@gm...>
>>
>> I've updated Golly's support for the Margolus neighborhood. It still
>> only works through emulation but now you can write any kind of
>> Margolus transition rule table, including ones with N states,
>> variables and different symmetries. This is a powerful tool, as I hope
>> some examples show:
>> DLA.rle (in Patterns/Other-Rules/Margolus) shows the famous diffusion
>> limited aggregation.
>> Sand.rle shows MCell's simple sand-like CA.
>> Both of these have very simple rule tables, as you can see here:
>> http://code.google.com/p/ruletablerepository/wiki/MargolusNeighborhood
>>
>> Both of these Margolus rule table files were auto-converted to
>> Moore-nhood files using a new script: emulate-Margolus-table.py in
>> Python/Margolus. You can find the rule tables it started from and the
>> ones it produces in Rules/.
>>
>> The old script that converted from MCell's string representation is
>> now called convert-MCell-string.py and outputs Margolus rule tables,
>> which can then be converted using the emulation script above.
>>
>> The coding of the cells has changed slightly, to better support N
>> states, and so import.py and export.py have been updated to reflect
>> this.
>>
>> (In theory we could now hide the details of the emulation from the
>> user: they would interact with an N-state CA but behind the scenes it
>> would be running a 2N+1 CA. I'm not sure this is a good idea though.)
>>
>> I have plans to expand the rule table format in other similar
>> directions (that won't have Golly support for the time being, even
>> through emulation) so now is the time to influence where the format
>> goes.
>>
>> Thanks,
>>
>> Tim
>>
>> P.S. Calcyman: I see you uploaded a DLA rule table based on the
>> HPP.table - was that a response to my wiki page on Margolus, or just
>> coincidence?
>>
>> On Thu, Oct 23, 2008 at 11:07 PM, Tim Hutton <tim...@gm...> wrote:
>> > Hello,
>> >
>> > I've discovered that it's not too hard to emulate the Margolus
>> > neighborhood using a Moore neighborhood CA. See
>> > Patterns/Other-Rules/Margolus/BBM.rle for a demonstration of Fredkin's
>> > Billiard Ball Model, borrowed from MCell.
>> >
>> > Instead of two states we need five but that's a small price to pay.
>> > See Scripts/Python/Margolus/ for some helper scripts to compose your
>> > own rule tables and to set up the partitioning colors required.
>> >
>> > Here are the rules that MCell comes with:
>> > http://psoup.math.wisc.edu/mcell/rullex_marg.html
>> > Some of these can be ported straight over, others will need some
>> > thought. I'm still not sure how Sand works, since it seems to use more
>> > than two states.
>> >
>> > Enjoy,
>> >
>> > Tim
>> >
>> > --
>> > Tim Hutton - http://www.sq3.org.uk
>> >
>> > Take the Organic Builder challenge -
>> > http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/
>> >
>>
>>
>>
>> --
>> Tim Hutton - http://www.sq3.org.uk
>>
>> Take the Organic Builder challenge -
>> http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/
>>
>>
>> ------------------------------------------------------------------------------
>> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
>> is a gathering of tech-side developers & brand creativity professionals.
>> Meet
>> the minds behind Google Creative Lab, Visual Complexity, Processing, &
>> iPhoneDevCamp as they present alongside digital heavyweights like
>> Barbarian
>> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
>> _______________________________________________
>> Golly-test mailing list
>> Gol...@li...
>> https://lists.sourceforge.net/lists/listinfo/golly-test
>
>
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
> is a gathering of tech-side developers & brand creativity professionals.
> Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, &
> iPhoneDevCamp as they present alongside digital heavyweights like Barbarian
> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
> _______________________________________________
> Golly-test mailing list
> Gol...@li...
> https://lists.sourceforge.net/lists/listinfo/golly-test
>
>



-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/

> Alan, that rule does show some key characteristics that have kept me
> motivated by Generations 345/3/6 for the past seven months.
>
> It even backs up my current interpretation about the synergistic role of
> emergent order in accelerating the spread of deterministic chaos.
>
> Your rule is clearly richer in puffers, richer in spaceship forms and has a
> 3 cell wick stretcher seed which allows a space filler to emerge naturally.
>

Thanks


But because there cannot be large patterns with stable edges, it is unlikely
> to have mechanisms for seeding new active centres at a great distance.
>

Turn up the last number a/b/c the c in the rule to get it more stable and
less expanding or shrinking, you do loose some of the variation as a result
and there is no one value that is best, there is impressive amounts of
variation between all the possible values.


However it does have some intriguing forms, especially the slow orthogonal
> glider which can form into rakes with a 2 in 7 slope.
>

Yeah, the slowest tiny glider I have ever seen, forgot to put it in my demo
file before I released it.


Another test it passes for me is that a glancing head on collision between
> the commonest spaceships produces chaotic growth which by iteration
> 10,000 exhibits almost all the patterns that I noticed while running your
> hand crafted demo to 15,000.
>
> x = 8, y = 11, rule = 124567/34/6
> .2E$.2C$D2AD$B2AB$4A3$4.4A$4.B2AB$4.D2AD$5.2C!
>

x = 9, y = 6, rule = 124567/34/6
$2.2A$.A$7.A$6.A!

x = 10, y = 6, rule = 124567/34/6
$2.2A$.A$7.2A$6.A!


The population at 10,000 is 11,238,510 compared to 89,976 / 371,576 for the
> fastest growing seeds I have found / created under Generations 345/3/6 but
> it appears to iterate comparably fast, presumably because hashing works
> efficiently on the rapidly stabilised chaotic mazes.
>

I don't plan to upgrade my Golly until the next public release. Can you now
do "falling game of sand" type things (maybe at a zoom further away that
1:1). That game and it variations are amazing.


2009/5/27 Tim Hutton <tim...@gm...>

> I've updated Golly's support for the Margolus neighborhood. It still
> only works through emulation but now you can write any kind of
> Margolus transition rule table, including ones with N states,
> variables and different symmetries. This is a powerful tool, as I hope
> some examples show:
> DLA.rle (in Patterns/Other-Rules/Margolus) shows the famous diffusion
> limited aggregation.
> Sand.rle shows MCell's simple sand-like CA.
> Both of these have very simple rule tables, as you can see here:
> http://code.google.com/p/ruletablerepository/wiki/MargolusNeighborhood
>
> Both of these Margolus rule table files were auto-converted to
> Moore-nhood files using a new script: emulate-Margolus-table.py in
> Python/Margolus. You can find the rule tables it started from and the
> ones it produces in Rules/.
>
> The old script that converted from MCell's string representation is
> now called convert-MCell-string.py and outputs Margolus rule tables,
> which can then be converted using the emulation script above.
>
> The coding of the cells has changed slightly, to better support N
> states, and so import.py and export.py have been updated to reflect
> this.
>
> (In theory we could now hide the details of the emulation from the
> user: they would interact with an N-state CA but behind the scenes it
> would be running a 2N+1 CA. I'm not sure this is a good idea though.)
>
> I have plans to expand the rule table format in other similar
> directions (that won't have Golly support for the time being, even
> through emulation) so now is the time to influence where the format
> goes.
>
> Thanks,
>
> Tim
>
> P.S. Calcyman: I see you uploaded a DLA rule table based on the
> HPP.table - was that a response to my wiki page on Margolus, or just
> coincidence?
>
> On Thu, Oct 23, 2008 at 11:07 PM, Tim Hutton <tim...@gm...> wrote:
> > Hello,
> >
> > I've discovered that it's not too hard to emulate the Margolus
> > neighborhood using a Moore neighborhood CA. See
> > Patterns/Other-Rules/Margolus/BBM.rle for a demonstration of Fredkin's
> > Billiard Ball Model, borrowed from MCell.
> >
> > Instead of two states we need five but that's a small price to pay.
> > See Scripts/Python/Margolus/ for some helper scripts to compose your
> > own rule tables and to set up the partitioning colors required.
> >
> > Here are the rules that MCell comes with:
> > http://psoup.math.wisc.edu/mcell/rullex_marg.html
> > Some of these can be ported straight over, others will need some
> > thought. I'm still not sure how Sand works, since it seems to use more
> > than two states.
> >
> > Enjoy,
> >
> > Tim
> >
> > --
> > Tim Hutton - http://www.sq3.org.uk
> >
> > Take the Organic Builder challenge -
> > http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/
> >
>
>
>
> --
> Tim Hutton - http://www.sq3.org.uk
>
> Take the Organic Builder challenge -
> http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/
>
>
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
> is a gathering of tech-side developers & brand creativity professionals.
> Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, &
> iPhoneDevCamp as they present alongside digital heavyweights like Barbarian
> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
> _______________________________________________
> Golly-test mailing list
> Gol...@li...
> https://lists.sourceforge.net/lists/listinfo/golly-test
>

On Fri, May 22, 2009 at 3:46 PM, Andrew Trevorrow <an...@tr...> wrote:
>> Are there
>> many cases where a pattern needs a script to be run at the beginning?
>
> Maybe not a "need", but I can easily think of a number of cases
> where it would be very handy:
>
> - For those huge replicators with long tapes, a script could zoom into
> the active region at a sensible scale rather than show a full view.
>
> - For some hash-friendly patterns like the metacell examples, a script
> could set the step size to a sensible value like 8^4 or whatever.
>
> - For the WireWorld/Langtons-ant.mc file, a script could do all the
> viewing steps I recommend in the file's comments: change the scale to
> 1:1, set the base step to 28, and change the color of state 3 to a
> dark brown.
>
> A couple of people on this list (Brice and Dave, from memory) have
> requested some sort of "container" format that allows various
> pattern-specific settings.  Using a zip file with pattern+script makes
> this easy to achieve.  It would be a shame not to take advantage of
> Golly's powerful scripting capabilities.

I can see the benefit. Could we have that the script runs
automatically (if desired) only if it is called 'autorun.py/.pl' (or
something) and lies in the root folder of the zip? That would make it
clearer, and avoid having scripts run that weren't intended to be.

And mentioning folders within zips reminds me that that's another
thing to think about - the UI would need to allow browsing through the
different folders, ideally.

-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/

Alan, that rule does show some key characteristics that have kept me  
motivated by Generations 345/3/6 for the past seven months.

It even backs up my current interpretation about the synergistic role  
of emergent order in accelerating the spread of deterministic chaos.

Your rule is clearly richer in puffers, richer in spaceship forms and  
has a 3 cell wick stretcher seed which allows a space filler to emerge  
naturally.

But because there cannot be large patterns with stable edges, it is  
unlikely to have mechanisms for seeding new active centres at a great  
distance.

However it does have some intriguing forms, especially the slow  
orthogonal glider which can form into rakes with a 2 in 7 slope.

Another test it passes for me is that a glancing head on collision  
between the commonest spaceships produces chaotic growth which by  
iteration 10,000 exhibits almost all the patterns that I noticed while  
running your hand crafted demo to 15,000.

x = 8, y = 11, rule = 124567/34/6
.2E$.2C$D2AD$B2AB$4A3$4.4A$4.B2AB$4.D2AD$5.2C!

The population at 10,000 is 11,238,510 compared to 89,976 / 371,576  
for the fastest growing seeds I have found / created under Generations  
345/3/6 but it appears to iterate comparably fast, presumably because  
hashing works efficiently on the rapidly stabilised chaotic mazes.


On 27/05/2009, at 5:41 AM, Alan Tennant wrote:

> Remove the 2x1 block on the bottom left to get it rolling, the other  
> things are just other things that also happen to work in the rule. I  
> know there is someone here who is really into this sort of thing and  
> had a really inefficient version of this rule.
> <Alan124567s34s6.rle>



Tony Smith
Complex Systems Analyst
Meme Media
Melbourne, Australia
http://www.meme.com.au/

I've updated Golly's support for the Margolus neighborhood. It still
only works through emulation but now you can write any kind of
Margolus transition rule table, including ones with N states,
variables and different symmetries. This is a powerful tool, as I hope
some examples show:
DLA.rle (in Patterns/Other-Rules/Margolus) shows the famous diffusion
limited aggregation.
Sand.rle shows MCell's simple sand-like CA.
Both of these have very simple rule tables, as you can see here:
http://code.google.com/p/ruletablerepository/wiki/MargolusNeighborhood

Both of these Margolus rule table files were auto-converted to
Moore-nhood files using a new script: emulate-Margolus-table.py in
Python/Margolus. You can find the rule tables it started from and the
ones it produces in Rules/.

The old script that converted from MCell's string representation is
now called convert-MCell-string.py and outputs Margolus rule tables,
which can then be converted using the emulation script above.

The coding of the cells has changed slightly, to better support N
states, and so import.py and export.py have been updated to reflect
this.

(In theory we could now hide the details of the emulation from the
user: they would interact with an N-state CA but behind the scenes it
would be running a 2N+1 CA. I'm not sure this is a good idea though.)

I have plans to expand the rule table format in other similar
directions (that won't have Golly support for the time being, even
through emulation) so now is the time to influence where the format
goes.

Thanks,

Tim

P.S. Calcyman: I see you uploaded a DLA rule table based on the
HPP.table - was that a response to my wiki page on Margolus, or just
coincidence?

On Thu, Oct 23, 2008 at 11:07 PM, Tim Hutton <tim...@gm...> wrote:
> Hello,
>
> I've discovered that it's not too hard to emulate the Margolus
> neighborhood using a Moore neighborhood CA. See
> Patterns/Other-Rules/Margolus/BBM.rle for a demonstration of Fredkin's
> Billiard Ball Model, borrowed from MCell.
>
> Instead of two states we need five but that's a small price to pay.
> See Scripts/Python/Margolus/ for some helper scripts to compose your
> own rule tables and to set up the partitioning colors required.
>
> Here are the rules that MCell comes with:
> http://psoup.math.wisc.edu/mcell/rullex_marg.html
> Some of these can be ported straight over, others will need some
> thought. I'm still not sure how Sand works, since it seems to use more
> than two states.
>
> Enjoy,
>
> Tim
>
> --
> Tim Hutton - http://www.sq3.org.uk
>
> Take the Organic Builder challenge -
> http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/
>



-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/

Alan Tennant wrote:

> Self replicating generations rule

What am I missing?  The pattern that you supplied, and most large random starting patterns, produce many different puffers, wickstretchers, and breeders, but I don't see anything that looks like self-replication.

Dean Hickerson


      

>  ... I just made an experimental change to the alphabetical archive 
>pages: now beside the pattern names, a little wiki logo that links 
>to the corresponding wiki page appears. The miniature wiki logo is 
>located at 
><http://www.conwaylife.com/w.png" rel="nofollow">http://www.conwaylife.com/w.png>http://www.conwaylife.com/w.png   I 
>tried just using <img src="w.png"> ...

For those img links to work I'd have to search the contents of the
downloaded html file and download any img files separately.
That opens a can of worms I'd rather not have to deal with at
the moment -- maybe in a future version.

>  My concern is that the pages that contain this new feature load 
>*extremely* slowly on my computer (only via the Golly viewer - not 
>in a web browser), while the other pages (the categorical listings) 
>that don't have those images on them still load normally. I imagine 
>that this is happening because the Golly browser is actually 
>querying the <http://conwaylife.com>conwaylife.com server every time 
>it displays that image.

Yep, that is what is happening, but I'm puzzled why it is so slow,
given the tiny size of the .png file.

>  Is there something that I can do to fix this? Should I remove the 
>images and/or the wiki links?

Best to remove all img links for now.

Andrew

> It would be nice to include a normal link to the main LifeWiki page,
> maybe by changing the 1st paragraph to:

> The <a href="http://www.conwaylife.com/wiki/index.php?title=Main_Page
">LifeWiki</a>
> pattern archive contains ...

> And maybe add a link somewhere to the ConwayLife.com forums page as well.
Thanks, that's done now. Also, I just made an experimental change to the
alphabetical archive pages: now beside the pattern names, a little wiki logo
that links to the corresponding wiki page appears. The miniature wiki logo
is located at http://www.conwaylife.com/w.png   I tried just using <img
src="w.png">, but that gave a broken image, so instead I had to use <img
src="." rel="nofollow">http://www.conwaylife.com/w.png">.

My concern is that the pages that contain this new feature load *extremely*
slowly on my computer (only via the Golly viewer - not in a web browser),
while the other pages (the categorical listings) that don't have those
images on them still load normally. I imagine that this is happening because
the Golly browser is actually querying the conwaylife.com server every time
it displays that image. Is there something that I can do to fix this? Should
I remove the images and/or the wiki links?

- Nathaniel Johnston


On Sat, May 23, 2009 at 7:01 PM, Andrew Trevorrow <an...@tr...>wrote:

> > Thanks, the online archive seems to work quite nicely. I've updated the
> LifeWiki archive file so that it now contains links to a full list, lists by
> first letter of the pattern, and (this part is a work in progress) lists by
> category. ...
>
> Looks great -- thanks!
>
> > Any other feedback and suggestions are of course welcome.
>
> It would be nice to include a normal link to the main LifeWiki page,
> maybe by changing the 1st paragraph to:
>
> The <a href="http://www.conwaylife.com/wiki/index.php?title=Main_Page
> ">LifeWiki</a>
> pattern archive contains ...
>
> And maybe add a link somewhere to the ConwayLife.com forums page as well.
>
> Andrew
>
>
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT
> is a gathering of tech-side developers & brand creativity professionals.
> Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, &
> iPhoneDevCamp asthey present alongside digital heavyweights like Barbarian
> Group, R/GA, & Big Spaceship. http://www.creativitycat.com
> _______________________________________________
> Golly-test mailing list
> Gol...@li...
> https://lists.sourceforge.net/lists/listinfo/golly-test
>

> Oh, one slight problem though - the default path doesn't exist, on the
> default Windows and Linux installs at least. The Rules folder hasn't
> been created.
> e.g. /home/Tim/.golly/Rules/ (on Linux)
> 
> This means that the script breaks. (Of course the user can change the
> path in the settings dialog, or create the folder, but this isn't an
> ideal solution as many of them won't know what went wrong.)

Good point.  I'll make sure Golly creates that folder if it doesn't
exist (simplest just to do it at start up time, after reading the
prefs file).

Andrew

> Thanks, the online archive seems to work quite nicely. I've updated the LifeWiki archive file so that it now contains links to a full list, lists by first letter of the pattern, and (this part is a work in progress) lists by category. ...

Looks great -- thanks!

> Any other feedback and suggestions are of course welcome.

It would be nice to include a normal link to the main LifeWiki page,
maybe by changing the 1st paragraph to:

The <a href="LifeWiki" rel="nofollow">http://www.conwaylife.com/wiki/index.php?title=Main_Page">LifeWiki</a>
pattern archive contains ...

And maybe add a link somewhere to the ConwayLife.com forums page as well.

Andrew

> Actually, I have one bug to report with the online archive -- if you currently have a pattern running, you can't download files or browse around the archive. Clicking on any link gives you "Could not download file (error 1)". It works fine if you press the stop button and then try the link again.

Good catch.  It seems that downloading is only possible from the main
idle event loop, so I'll add a suitable warning.

Andrew

Thanks Andrew, I'll make sure to use this approach in future.

Oh, one slight problem though - the default path doesn't exist, on the
default Windows and Linux installs at least. The Rules folder hasn't
been created.
e.g. /home/Tim/.golly/Rules/ (on Linux)

This means that the script breaks. (Of course the user can change the
path in the settings dialog, or create the folder, but this isn't an
ideal solution as many of them won't know what went wrong.)

Should that folder exist? i.e. Should we get Golly to create it if it
doesn't exist?

Or should I be testing for the existence of the rulesdir() folder in my scripts?

On Fri, May 15, 2009 at 4:57 AM, Andrew Trevorrow <an...@tr...> wrote:
> I noticed Tim added a new script with this line:
>
> f=open('../../Rules/Fredkin-mod-n.table', 'w')
>
> Writing new table files in the supplied Rules folder is not such a
> good idea because Golly might be installed in a read-only directory.
> So I've added a new rulesdir() command which returns the path to
> the user's rules folder (as set in Prefs > Control) and changed
> the script to do:
>
> f=open(golly.rulesdir()+'Fredkin-mod-n.table', 'w')
>
> PS. People who install Golly in a writable directory and who
> don't want 2 different rules folders can always set their rules
> folder to the supplied Rules folder.
>
> Andrew
>
> ------------------------------------------------------------------------------
> Crystal Reports - New Free Runtime and 30 Day Trial
> Check out the new simplified licensing option that enables
> unlimited royalty-free distribution of the report engine
> for externally facing server and web deployment.
> http://p.sf.net/sfu/businessobjects
> _______________________________________________
> Golly-test mailing list
> Gol...@li...
> https://lists.sourceforge.net/lists/listinfo/golly-test
>



-- 
Tim Hutton - http://www.sq3.org.uk

Take the Organic Builder challenge -
http://www.sq3.org.uk/Evolution/Squirm3/OrganicBuilder/