Pattern viewer for forum threads

[rowett]

There are definitely still issues outside of Margolus from what I can see.

Fixed in build 1272, thanks.

[muzik]

There are no history trails here when loaded, but if we change the theme to Blues, evolve for more generations, and then switch back to Mono, there obviously are generations as can be seen from the stacked layers (although they disappear immediately if we start playback again). Also, in this state, Help > Info > Cells will say the basic shader is in use even though it presumably isn't.

Code: Select all

x = 7, y = 5, rule = B3/S23
3b2o$bo4bo$o$o5bo$6o!
[[ ZOOM 4 STARTFROM 30 LAYERS 10 GRID THEME Mono ]]

Line breaks appear to be ignored in the comments menu. Also, I don't know if the lines for #N and #O should be moved from Info > Pattern to the comments menu, or should remain in both places.

Code: Select all

#C generate a random map which overrides the defined rule below
#N Acorn
#O Charles Corderman
#C A methuselah with lifespan 5206.
#C http://www.conwaylife.com/wiki/index.php?title=Acorn
x = 7, y = 3, rule = B3/S23
bo5b$3bo3b$2o2b3o!

Since a message is displayed when cells are killed due to impact with the outer grid edge, should another be displayed for KILLGLIDERS?

How can I embed a viewer window in a wiki page which is immediately interactable in an "expanded" state, instead of starting out as a thumbnail?

I believe this should show an error, since you're changing a color for something that under these circumstances does not exist:

Code: Select all

x = 4, y = 4, rule = B3/S23
2b2o$3bo$o$2o!
[[ GRIDMAJOR 0 COLOR GRIDMAJOR Blue ]]

Should Help > Themes be displaying a value for Dead for themes such as Mono if it's identical to Background, and Help > Info > Cells doesn't display it?

Can icon data be moved to Help > Info > Cells where appropriate?

In this pattern, alive cells are placed in the exact same positions as the grid boundaries. If we delete all the visible living cells at T 0, we still have 17 alive cells. Using Select All will select these despite them not being visible. Also, if we turn on the Neighbor shader, living cells inside of bounds will be colored as though those boundary cells are alive (so I'm wondering if we could make state 254 be an "alive" boundary cell to fix the issue where cells aren't colored correctly across connected boundaries). When we go forward a generation, however, all of these cells immediately die.

Code: Select all

x = 10, y = 10, rule = M0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15:T10,10
$b10o$b10o$b10o$b10o$b10o$b10o$b10o$b10o$b10o$b10o!
[[ SHOWGENSTATS ]]

It's possible to start drawing if you click on the bottom boundary or the right boundary, which isn't true outside of Margolus. Compare:

Code: Select all

x = 10, y = 10, rule = M0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15:T10,10
10b$10b$10b$10b$10b$10b$10b$10b$10b$10b!

Code: Select all

x = 10, y = 10, rule = B/S01234V:T10,10
10b$10b$10b$10b$10b$10b$10b$10b$10b$10b!

Can base layer darkening be turned off when in the Select menu?

Code: Select all

x = 5, y = 4, rule = B3/S23
bo2bo$o$o3bo$4o!
[[ LAYERS 10 ]]

And finally, would there be any performance benefit to introducing a shader that doesn't have cell ageing for alive cells or dead cells, but still accounts for a history trail using only one state (as is done for the Catagolue theme, or any theme if we use AGESTATES 0 HISTORYSTATES 1)?

[rowett]

There are no history trails here when loaded, but if we change the theme to Blues, evolve for more generations, and then switch back to Mono, there obviously are generations as can be seen from the stacked layers

This is known and is part of the Shaders work to be completed.

Line breaks appear to be ignored in the comments menu. Also, I don't know if the lines for #N and #O should be moved from Info > Pattern to the comments menu, or should remain in both places.

Fixed in build 1273.

Note: Since the bot protections were added to the forums it can take up to 4 hours for a new build to appear. The CDN is caching Javascript files. Hard refresh in the browser makes no difference as it just gets the latest from the CDN, and the CDN only refreshes from the source when the cache expires.

Since a message is displayed when cells are killed due to impact with the outer grid edge, should another be displayed for KILLGLIDERS?

No. KILLGLIDERS is intentional. A glider hitting the grid boundary likely isn't and so is notable.

I believe this should show an error, since you're changing a color for something that under these circumstances does not exist

In this case it's better to silently ignore it.

In this pattern, alive cells are placed in the exact same positions as the grid boundaries.

Margolus issues are known - no need to report any more at the moment.

Can base layer darkening be turned off when in the Select menu?

Yes, done.

And finally, would there be any performance benefit to introducing a shader that doesn't have cell ageing for alive cells or dead cells, but still accounts for a history trail using only one state?

No.

[rowett]

How can I embed a viewer window in a wiki page which is immediately interactable in an "expanded" state, instead of starting out as a thumbnail?

See LifeViewer.

[rowett]

Steps to reproduce (build 1260):

1. Click "Show in viewer".
2. Enter selection mode (F4), and select the "top half" of the pattern's generation 0 (select all alive cells that are above the horizontal line dividing the pattern in half).
3. ...

Fixed in build 1275, thanks.

[muzik]

https://lazyslug.com/lifeviewer/pro/ doesn't appear to work correctly on Falkon, producing an error and defaulting to Standard. I'm not sure if there's some sort of anti-fingerprinting measure getting in the way, but more privacy-focused browsers such as Mullvad and Librewolf appear to work fine in this aspect.

wasmerror.png (8.41 KiB) Viewed 676 times

I believe this should show an error, since you're changing a color for something that under these circumstances does not exist

In this case it's better to silently ignore it.

As I see it, it's a color that can be customized in a different context, but cannot be in this specific case. LifeViewer already handles this for COLOR BOUNDED when there's no bounded grid specified, as well as for cell states that aren't present in the current rule, so (at least as I see it) this should follow that sort of initiative to be consistent.

This discrepancy is probably known and expected due to shader handling differences, but documenting it here for completeness:

Code: Select all

x = 3, y = 1, rule = B3/S23
3o!
[[ THEME Mono LAYERS 10 X 5 Y 5 AUTOSTART ]]

Code: Select all

x = 3, y = 1, rule = R1,C2,S2-3,B3
3o!
[[ THEME Mono LAYERS 10 X 5 Y 5 AUTOSTART ]]

How exactly is LifeViewer embedded on the LifeWiki and how can it be embedded on another wiki? We're trying to get this to work on minecraft.wiki (see here and here), but the test case just appears to be generating a LifeViewer-sized area of blank space on most browsers.

[rowett]

https://lazyslug.com/lifeviewer/pro/ doesn't appear to work correctly on Falkon, producing an error and defaulting to Standard. I'm not sure if there's some sort of anti-fingerprinting measure getting in the way, but more privacy-focused browsers such as Mullvad and Librewolf appear to work fine in this aspect.

Glad it defaulted to Standard. At some stage I'll need to test it when it's hosted here at conwaylife.com.

How exactly is LifeViewer embedded on the LifeWiki and how can it be embedded on another wiki? We're trying to get this to work on minecraft.wiki (see here and here), but the test case just appears to be generating a LifeViewer-sized area of blank space on most browsers.

Not entirely sure - I think Nathaniel did the integration.

Looking at your test case page it had the following line:
If this doesn't appear embedded, check if it appears here: https://tools.minecraft.wiki/static/tools/lifeviewer/

Following that link also has a blank space where the viewer should be. Looking at the HTML it says:

Code: Select all

<script src="./lv-viewer.js"></script>

But the server doesn't seem to have the lv-viewer.js file at that location so LifeViewer is never loaded.

[muzik]

Looking at your test case page it had the following line:
If this doesn't appear embedded, check if it appears here: https://tools.minecraft.wiki/static/tools/lifeviewer/

Following that link also has a blank space where the viewer should be. Looking at the HTML it says:

Code: Select all

<script src="./lv-viewer.js"></script>

But the server doesn't seem to have the lv-viewer.js file at that location so LifeViewer is never loaded.

It used to embed fine before the second PR, but only on browsers which were Firefox and forks thereof. Chromium browsers and Safari didn't work.

[rowett]

It used to embed fine before the second PR, but only on browsers which were Firefox and forks thereof. Chromium browsers and Safari didn't work.

Well from my brief look:

1. the test page doesn't include the LifeViewer plugin at all
2. as mentioned above, the page linked from there tries to include LifeViewer but it's not at the requested location

[muzik]

Asked Nathaniel about this on Discord, so I'm not exactly sure as to where to start; do either of you have any logs from conversations at that time that could carry relevant implementation details?

IMG_7020.jpeg (161.17 KiB) Viewed 613 times

IMG_7021.jpeg (311.62 KiB) Viewed 613 times

In other news: THUMBLAUNCH appears to not be erroring correctly with invalid pattern data. Also, interacting with the first of the following two viewers opens a popup window whose dimensions are much smaller than the enforced default, so it seems like something may have been forgotten with the recent changes and there are still parts of the codebase that wrongly believe that the minimum dimensions are the old ones and should be updated accordingly.

Code: Select all

x = 3, y = 3, rule = invalid-rule-name
o$obo$2o!
[[ THUMBLAUNCH ]]

Code: Select all

![[ THUMBLAUNCH ]]

[muzik]

Progress update: we're now at a state where the viewer embed appears and is interactable. However, rules simply do not work. Attempting to load Langtons-Ant inline gets us B3/S23 that iterates at 2 gps, but ages cells all the way to T=64 as though it were running at 1x. If we instead set it to run a rule such as B3/S5, it insists that there's an illegal character in the survival specification.

I'm not sure if some weird bugs are being exposed in LifeViewer in the process of embedding it, or if the specific version being embedded is a broken copy.

I've tried saving this, but it doesn't show up embedded on web.archive.org: https://web.archive.org/web/20250308011 ... ki:Sandbox

EDIT: We're now past this point and have something more functional.

EDIT2: assuming subsequent fixes don't stop this, the following should display the broken behaviour - try drawing, watching the evolution, and then attempt to reset: https://minecraft.wiki/w/Minecraft_Wiki ... id=2889020

[rowett]

However, rules simply do not work

They won't. LifeViewer needs the rules to be in the same domain (i.e. it's not allowed to access LifeWiki from that website). Your options are:

1. Put the rule after the RLE in the pattern
2. Create a local repository on your website and adjust LifeViewer meta settings accordingly
3. Talk to Nathaniel about allowing access

assuming subsequent fixes don't stop this, the following should display the broken behaviour

This is likely caused by a bug just fixed in build 1276.

[rowett]

Here's another Identify test case for potential optimization. Results on iPad safari, build 1265:

WASM off: 30.5s / 46.3s, no strict volatility
WASM on: 4.5s / 8.9s

Optimized.

On my desktop machine [Win 11, Chrome]:

• WASM off: 15.3s / 16.2s, no strict volatility
  WASM on: 2.7s/ 5.3s

[muzik]

LifeViewer is now used on the Minecraft Wiki to demonstrate Langton's Ant:

https://minecraft.wiki/w/An_Ant#Usage

[rowett]

LifeViewer is now used on the Minecraft Wiki to demonstrate Langton's Ant:

Excellent - well done! That wins the prize as the most unlikely place I thought I'd find LifeViewer (so far...)

[LuveelVoom]

LifeViewer is now used on the Minecraft Wiki to demonstrate Langton's Ant:

https://minecraft.wiki/w/An_Ant#Usage

Great! Will it be used on the fluids page and maybe on the pages about pre-1.18 world gen?

[muzik]

LifeViewer is now used on the Minecraft Wiki to demonstrate Langton's Ant:

https://minecraft.wiki/w/An_Ant#Usage

Great! Will it be used on the fluids page and maybe on the pages about pre-1.18 world gen?

I have an old demo for fluids that may come in useful. Other things I've considered include lighting and leaf decay. What aspects of pre-1.18 world generation would be relevant to cover?

[LuveelVoom]

I believe they used cellular automata, did they not? How would leaf decay be done with randomness?

[muzik]

I believe they used cellular automata, did they not? How would leaf decay be done with randomness?

It's listed here: https://minecraft.wiki/w/Tick#Random_tick

Eligibility is radius-based and can be implemented as a simple rule.

assuming subsequent fixes don't stop this, the following should display the broken behaviour

This is likely caused by a bug just fixed in build 1276.

We now get an "unsupported rule name" error which is probably correct.

I had to request that the viewer version be manually updated to see this. How can this be automated, if they decide this is something they want to do?

[muzik]

In the interests of embedding LifeViewer in other sites and having it fit better: is there a way to translate the text in LifeViewer into other languages, and is there a way to change the colours of the top bar in the popup as well as the close button (either via CSS or something, or by using script commands like with other UI elements)?

On certain screen sizes such as phones, embedded viewers may not fit on screen correctly, with important buttons being obscured. Is there a way around this so that viewers can be used on such devices as intended?

IMG_7084.jpeg (352.18 KiB) Viewed 407 times

Comment data appears to persist into subsequent pattern files even if they don't have any comments - try deleting the first line in https://lazyslug.com/lifeview/plugin/viewer.html, pressing View, and then entering the comments menu.

Yet to test this on other devices, but at least on an iPad, using Cut instead of Copy on the clipboard contents text box will result in subsequent in-viewer cut/copy options not updating the contents of the box.

In the frequency table, can the rightmost column (currently empty) be used for what percentage of the non-0 frequency cells an oscillator comprises? For cases like the following, 0 dominates, making almost all of the other percentages extremely low.

Code: Select all

x = 290, y = 263, rule = B3/S23
276b2o$274bo7bo4bo$275bobo2b2ob4ob2o$276bo5bo4bo257$2bo4bo$2ob4ob2o$2bo4bo!
[[ AUTOIDENTIFY ]]

Also, in Help > Info > Identify, there are listed values for both 0 and Back in the frequencies table. Back is effectively a duplicate of existing information and can probably be removed here. It would only be relevant in cases like this, where there aren't any frequency-0 cells in the bounding box.

Code: Select all

x = 7, y = 5, rule = B2in3/S123a
bo$o4bo$bo3bo$bo4bo$5bo!
[[ AUTOIDENTIFY ]]

[muzik]

Switch to another theme such as the Mono theme, and the loop multiplier won't change, even though it should to accommodate the major grid lines. The loop duration would be expected to change to avoid the jumping grid lines here.

Code: Select all

x = 3, y = 3, rule = B3/S23
bo$o$3o!
[[ AUTOSTART THEME Margolus GRID TRACKLOOP 4 -1/4 1/4 GPS 5 ]]

Will the maximum Pro Identify generation be increased beyond 10 million, so that oscillators like the following can be identified, as shown in this post?

Code: Select all

x = 8, y = 7, rule = M0,1,2,3,4,10,6,11,8,9,5,13,12,14,7,15
5bo$5b2o$4bo2bo$5bobo$bob3obo$5bobo$4bo!

[rowett]

In the interests of embedding LifeViewer in other sites and having it fit better: is there a way to translate the text in LifeViewer into other languages

Not currently. It's probably a chunk of work. I'd need to:

1. Identify all of the places that output text (UI controls, tooltips, error messages, notifications, Help text, etc.)
2. Replace all of the messages with message ids
3. Create a language module from where I can lookup a particular id in a particular language
4. Get people to create language packs

is there a way to change the colours of the top bar in the popup as well as the close button (either via CSS or something, or by using script commands like with other UI elements)?

Yes from build 1279:

• COLOR TITLEFOREGROUND R G B
• COLOR TITLEBACKGROUND R G B
• COLOR CLOSEFOREGROUND R G B
• COLOR CLOSEBACKGROUND R G B

On certain screen sizes such as phones, embedded viewers may not fit on screen correctly, with important buttons being obscured. Is there a way around this so that viewers can be used on such devices as intended?

Not currently, but this is on the backlog.

Comment data appears to persist into subsequent pattern files even if they don't have any comments

Fixed, thanks.

Yet to test this on other devices, but at least on an iPad, using Cut instead of Copy on the clipboard contents text box will result in subsequent in-viewer cut/copy options not updating the contents of the box.

Fixed, thanks.

In the frequency table, can the rightmost column (currently empty) be used for what percentage of the non-0 frequency cells an oscillator comprises?

It seems a bit arbitrary for this particular case. In other maps it would be confusing.

Also, in Help > Info > Identify, there are listed values for both 0 and Back in the frequencies table.

Back now appears with the Cell Period data if used.

[rowett]

Will the maximum Pro Identify generation be increased beyond 10 million, so that oscillators like the following can be identified, as shown in this post?

Code: Select all

x = 8, y = 7, rule = M0,1,2,3,4,10,6,11,8,9,5,13,12,14,7,15
5bo$5b2o$4bo2bo$5bobo$bob3obo$5bobo$4bo!

That's a LifeViewer Ultra feature. It's embarrassingly expensive.

[muzik]

is there a way to change the colours of the top bar in the popup as well as the close button (either via CSS or something, or by using script commands like with other UI elements)?

Yes from build 1279:

• COLOR TITLEFOREGROUND R G B
• COLOR TITLEBACKGROUND R G B
• COLOR CLOSEFOREGROUND R G B
• COLOR CLOSEBACKGROUND R G B

The values for these don't appear to be listed anywhere in Help > Info, regardless of whether they're customized or not. I'd most likely expect them to be in the UI section.

Code: Select all

x = 191, y = 144, rule = B3aiky4ace5aejk6in/S2en3cij4aciw5-cn6aei7e
12$122bo$123bo5$112bo13b3o$110b5o11bobo$108b4ob4o11bo$106b3obo3bob3o$
104b4obo4bo2bo$102b3obobo5bobo$103bobo$102b3obo$101bobobo$104b2o$102b
3o2$112bo$110b5o$108b4ob4o18bo$106b3obo3bob3o15b2o$104b4o2bo3bo2b4o14b
2o$102b3obo7bobobob3o11bobo$100b4o2bo8b2ob2ob4o8bob3o$99b2ob2o2bo8bo3b
o2bob2o6bobob2o$98b2obo17bobobob2o6b3ob2o$98bob2o24bo5bob3obo$97b5o2bo
20b3o7b5o$97b4ob3o21b2o4b3ob4o$96b2ob4o9bo12bob2o5b4ob2o$96bob3o9b5o
13bo7b3obo$95b4o9b4ob4o8b5o8b4o$95b3o9b2ob2ob2ob2o9b3o9b3o$94b2o10b2ob
2o3bo2b2o10b2o10b2o$94bo11bo2bo4b3obo11bo11bo$93b5o7b4o8b3o9b3o9b3o$
93b3o9b3o7b2obo11bo11bo$92b2o10b2o8b2ob2o10b2o10b2o$81bo10bo11bobo8b3o
9b3o10b2o$79bob3o7b3o9b3o10bo10b3o9b3o$80bo11bo11bobo19bobo9bobo$80b2o
10b2o10b2o21b2o10b2o$81b3o9b3o9b2o19b2o9b3o$81bob3o7b3o9b3o16b4o7b5o$
84bo9bobo9bo15b3obo6b2o3bo$94b2o10b2o2bo11bo2b2o7b2ob2o$95b2o10b2ob2o
9b2ob2o10b2o$95b4o9b4ob2o3bo2b4o8b5o$96bo13b3ob2ob2ob3o13bo$96b2o14b4o
b4o8b2o4b2o$97b3o14b5o10bo3b3o$97b3o16bo11b2o3b3o$98bobo2b3o22bob3obo$
98b2o5bo21b5ob2o$99b2ob3obo20b4ob2o$100b4obo20b2ob4o$102b3obobo17bob3o
$104b4obo13bob4o$106b3ob2o2bo3bo2b2ob3o$108b4o2bo3bo2b4o$110b3obo3bob
3o$112b4ob4o$114b5o$116bo!
[[ MAXGRIDSIZE 9 ]]

Code: Select all

x = 191, y = 144, rule = B3aiky4ace5aejk6in/S2en3cij4aciw5-cn6aei7e
12$122bo$123bo5$112bo13b3o$110b5o11bobo$108b4ob4o11bo$106b3obo3bob3o$
104b4obo4bo2bo$102b3obobo5bobo$103bobo$102b3obo$101bobobo$104b2o$102b
3o2$112bo$110b5o$108b4ob4o18bo$106b3obo3bob3o15b2o$104b4o2bo3bo2b4o14b
2o$102b3obo7bobobob3o11bobo$100b4o2bo8b2ob2ob4o8bob3o$99b2ob2o2bo8bo3b
o2bob2o6bobob2o$98b2obo17bobobob2o6b3ob2o$98bob2o24bo5bob3obo$97b5o2bo
20b3o7b5o$97b4ob3o21b2o4b3ob4o$96b2ob4o9bo12bob2o5b4ob2o$96bob3o9b5o
13bo7b3obo$95b4o9b4ob4o8b5o8b4o$95b3o9b2ob2ob2ob2o9b3o9b3o$94b2o10b2ob
2o3bo2b2o10b2o10b2o$94bo11bo2bo4b3obo11bo11bo$93b5o7b4o8b3o9b3o9b3o$
93b3o9b3o7b2obo11bo11bo$92b2o10b2o8b2ob2o10b2o10b2o$81bo10bo11bobo8b3o
9b3o10b2o$79bob3o7b3o9b3o10bo10b3o9b3o$80bo11bo11bobo19bobo9bobo$80b2o
10b2o10b2o21b2o10b2o$81b3o9b3o9b2o19b2o9b3o$81bob3o7b3o9b3o16b4o7b5o$
84bo9bobo9bo15b3obo6b2o3bo$94b2o10b2o2bo11bo2b2o7b2ob2o$95b2o10b2ob2o
9b2ob2o10b2o$95b4o9b4ob2o3bo2b4o8b5o$96bo13b3ob2ob2ob3o13bo$96b2o14b4o
b4o8b2o4b2o$97b3o14b5o10bo3b3o$97b3o16bo11b2o3b3o$98bobo2b3o22bob3obo$
98b2o5bo21b5ob2o$99b2ob3obo20b4ob2o$100b4obo20b2ob4o$102b3obobo17bob3o
$104b4obo13bob4o$106b3ob2o2bo3bo2b2ob3o$108b4o2bo3bo2b4o$110b3obo3bob
3o$112b4ob4o$114b5o$116bo!
[[
COLOR BACKGROUND Purple
COLOR DEAD Purple
COLOR DEADRAMP Purple
COLOR ALIVE Pink
COLOR ALIVERAMP Pink
COLOR GRID Pink
COLOR GRIDMAJOR Pink
COLOR BOUNDARY Pink
COLOR SELECT Pink
COLOR PASTE Pink
COLOR SELECTED Pink
COLOR GRAPHBG Pink
COLOR GRAPHAXIS Pink
COLOR GRAPHALIVE Pink
COLOR GRAPHBIRTH Pink
COLOR GRAPHDEATH Pink
COLOR UIFOREGROUND Pink
COLOR UIBACKGROUND Pink
COLOR UIHIGHLIGHT Pink
COLOR UISELECT Pink
COLOR UILOCKED Pink
COLOR UIBORDER Pink
COLOR TEXT Pink
COLOR ERROR Pink
COLOR HELP Pink
COLOR STARS Pink
COLOR TITLEBACKGROUND Pink
COLOR TITLEFOREGROUND Pink
COLOR CLOSEBACKGROUND Pink
COLOR CLOSEFOREGROUND Pink
MAXGRIDSIZE 9 ]]

The close button in Settings also appears to remain hardcoded in color, which I don't think should happen: it should follow the newly-customizable close button color as per this post.

For completeness (and since I intended to mention this in the quoted post but forgot): viewer popup windows also display a drop shadow, which is most prominent at the bottom and right edges. There is also a very thin gray border around viewer popups. Could the colour of these two things also be made customizable?

It may also be desirable in some cases to not have a drop shadow at all on the popup - perhaps an alpha setting could be introduced when customizing colours (I'd prefer it to use ARGB or four values from 0 to 255, instead of a value from 0.0 to 1.0). This could also be extended to existing UI elements, since they have varying transparency levels.

[muzik]

Selections still don't appear to work correctly on bounded grids if CXRLE Pos is specified - they follow the cursor correctly, but do not represent the actual region that gets selected internally (Select All does, and therefore appears offset):

Code: Select all

#CXRLE Pos=0,0
x = 20, y = 20, rule = B3/S23:T20,20
20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o$20o
$20o$20o!

The X and Y commands used to position the camera at the top center of the grid. It now positions the camera at the top left. I do not know whether the old behaviour of the X and Y commands on the hexagonal grid is correct, or the current behaviour is.

And finally, if we use this theme from a while ago, one history trail cell appears to be missing at T=381: there are 11 deaths, but only 10 visible dead trail cells.

Code: Select all

x = 5, y = 5, rule = B2/S2H
o$2o$bo2$obobo!
[[ COLOR ALIVE 255 255 128 COLOR ALIVERAMP 255 255 128 HISTORYSTATES 2 COLOR BACKGROUND 0 0 0 COLOR DEAD 192 0 0 COLOR DEADRAMP 0 0 0 SHOWGENSTATS ZOOM 16 X -128 Y -256 STARTFROM 380 MAXGRIDSIZE 9 ]]

This specifies HISTORYSTATES 1, so I'd expect all dead cells to be the exact same. However, the two cells following this spaceship are not state 1: they appear higher on layers, and turning on State Number shows them to be state 63 and state 62. Should dead cells not instantly be state 1 if there's only meant to be one history state?

Code: Select all

x = 4, y = 2, rule = B2/S
o2bo$b2o!
[[ COLOR DEADRAMP White HISTORYSTATES 1 LAYERS 10 GRID STARTFROM 5 ]]

Likewise, if we set AGESTATES to 0, the bottom right cell here will be state 64, but the other three will be state 65. If we continue playing, everything reaches state 127. Shouldn't everything stay at state 64 if age states are disabled like this?

Code: Select all

x = 2, y = 2, rule = B3/S23
2o$o!
[[ AGESTATES 0 STARTFROM 1 ]]

This could work interestingly as a benchmark:

Here is a p669 "beyblade" oscillator on the hexagonal tiling:

Code: Select all

x = 42, y = 42, rule = R15,C10,S74-146,B111-148,N@018000001FF00000FFFC0001FFFC0007FFFE000FFFFE001FFFFE007FFFFF00FFFFFF00FFFFFE01FFFFFE03FFFFFE07FFFFFE07FFFFFC0FFFFFFC1FFFFFF83FFFFFF03FFFFFE07FFFFFE07FFFFFC07FFFFF807FFFFF00FFFFFF00FFFFFE007FFFF8007FFFF0007FFFE0003FFF80003FFF00000FF800000180H
8.3A2B$6.6A5B$4.9A7B$2.12A6B2C$2.8A7D4B3C$.6A2C6D5EB4C$.5A3C5D8E5C$.
4A4C4D7E3FE4C$5A4C4D5E7F3CD$4AB4C4D4E9F3C2D$4AB4C4D4E4F2G5FC3D$4A2B3C
5D3E3F8GF4D$4A2B3C6D3E3F8GF4D$.3A3B3C6D3E2F10G4D$.4A2B4C6D2E2F4G3H4G
3D$.4A3B5C5D2E2F3G5H3G2DE$2.4A2B7C4D2EF3G6H3G3E$2.4A3B8C3D2EF2G8HGD2E
$3.4A4B7C3ADE2G3H3I3HG3E$3.5A9B6A2G2H5I3HG3E$4.5A7B8AGH7I3H3E$5.5A5B
9AGH8I3H3E$6.5A2B12A2I6.2I2H3E$7.18AI8.2I2HE2F$8.16A10.3IH3F$10.11A
14.3I3F$11.10A15.2I3F$12.8A17.I4F$14.6A17.2I3F$17.2A18.2I3F$38.F3G$
38.4G$38.4G$38.4G$38.4G$37.3H2G$36.5H$35.6H$33.3I4H$30.7I2H$31.7I$32.
4I!

iPad Safari results, Pro build 1278:
WASM Engine off: 12.3s /14.8s
WASM Engine off, Use Rectangles: 11.4s / 13.9s
WASM Engine on: 5.7s / 6.7s
WASM Engine on, Use Rectangles: 5.3s / 6.3s

I'm not sure how much the renderer is impacting Identify performance, if at all.

It is possible to start making a selection when clicking out of bounds in the first of these two examples, but not the second:

Code: Select all

x = 1, y = 1, rule = B3/S23
!
[[ MAXGRIDSIZE 9 X -250 ZOOM 16 ]]

Code: Select all

x = 1, y = 1, rule = B3/S23:P480
!
[[ MAXGRIDSIZE 9 X -250 ZOOM 16 ]]

I still have inconsistent framerates when using LifeViewer on the forums, which impacts playback speed and (on iPad) usability. Is there any way to enforce 60gps at all when on iPad it can be slower for unknown reasons?