Catagolue Oddities

[drc]

Gameoflifeboy introduced me to a few weird things on catagolue, which I've combined with my own findings here:

• There used to be several still life categories in (like xs5) for which no objects existed. Instead of just displaying a normal error, it would show this, but they fixed it after I asked about it, so this is just a mock-up. [drc]

• Category z, which was a typo for zz, as seen in a few rules like b2i3678s34678. This was fixed in newer versions, but still remains. [apple bottom]

• Typo symmetries. [gameoflifeboy]

• If you go to any object categories from here or here, you will find they don't work. This is because of the percent symbol, and can be fixed by appending "25" after the percent sign of the object page. [drc]

• Although they have been moved, there are still C1 sample soups for LINEAR and PATHOLOGICAL. [drc]

• Broken snake game here. [gameoflifeboy]

• Linear growth that is just a GPSE with a little debris caused by a glider collision. [gameoflifeboy]

• Linear growth pattern haul, because catagolue didn't report them [gameoflifeboy]

• Weird catagories of hauls, 1 (red-orange), 4 (red), and 6 (blue) [drc]

• The zz_LINEAR is displayed on Adam's discovery page, along with the actual object category. [drc]

• Although technically not an oddity, APG is Adam's initials, although he says it stands for "Ash Pattern Generator" [drc]

• Apparently linear growths can contain minus signs. [muzik]

• Objects can have multiple b/s conditions, as seen here [muzik]

• Objects can have b9 and s9, as seen here [muzik]

• B/s conditions can be out of order, as seen here [drc]

• xp0 exists. [a for awesome]

• xp-1 gives a 500 server error.

• Adding letters v-z and 0 just right still retains the apgcode, as seen here [drc]

• Any pattern that doesn't fit into a named category is regarded as... a pattern

[muzik]

Minus-containing puffer: http://catagolue.appspot.com/object/yl1 ... 3dd/b38s23

S9 rule: https://catagolue.appspot.com/object/xp ... 4r4/b3s239

...Whatever the hell this is: https://catagolue.appspot.com/object/xp ... 2322323333

[drc]

Great! I'll add those to the master list. I'll also add the people who found them!

I remember posting that 2323322322323332 thing though.

[A for awesome]

This.

"This pattern is a pattern."

xp0's

catagolue.appspot.com/hashsoup works rather non-intuitively.

[biggiemac]

Much of this is just autogenerated from the URL input in a way one would hope, so it can hardly be said to be "on catagolue." Examples are putting nonstandard strings as the symmetry, the b/s, the apgcode. It might be worth addressing the hashsoup, which I agree parses very nonintuitively. It is nice that xp0 makes no checks on the behavior of the pattern, so it could be used to verify for a given apgcode that you're encoding the right pattern. I guess for a full public release one should make sure nobody could do damage by messing around with the URLs, but really if you are messing around with the URL to get to strange pages you shouldn't think it's bad when something harmlessly funny shows up.

The oddities are when there is an associated database entry for the autogenerated page, that you wouldn't expect. This includes comments on pages that shouldn't exist (like the "Any random thing can be an apgcode, apparently" page), hauls submitted to rule/symmetry combinations that are flawed (like the D4 +4 with no underscore or my 25% and 75% hauls), Adam's switch engine haul for compensating the (version 2.3?) snafu. It might be important to disable comments on pages without patterns, to prevent some bot from generating all possible xp0 pages and leaving a comment on each to flood the database with nodes. But then discussion on objects like the loafer wouldn't be possible.

The puffer notation issue is a tough one. The minus sign is a consequence of the method used to determine uniqueness. I don't think the puffer notation is very illuminating, at least all I understand is yl[period of growth]. The following two values don't make sense, and then I think what follows is an apgcode but I'm not sure. I don't believe there are plans to change it, but it could be nice to make it more understandable. Plus it is concerning that the protocol wasn't quite robust enough to handle a rogue glider in the GSPE + debris pattern, and also that it labelled the zz_LINEAR as it did.

[muzik]

I remember posting that 2323322322323332 thing though.

Pretty sure you did it in response to someone posting the 9 one. It think it was A for awesome but I could be wrong

[Apple Bottom]

Another oddity...

b36is2-i3eiq4ay has a "C1-2016-04-05-2016-04-05" symmetry. This one looks like a census backup, and in fact displays a message to this effect, but it is listed on the rule's list of symmetries. What this ultimately means only Calcyman knows.

• Weird catagories of hauls, 1 (red-orange), 4 (red), and 6 (blue) [drc]

To the best of my knowledge, 1 means "verified, but not yet uncommitted to the census", and 4 means "rejected" (failed chi-square test, or possibly some other check).

I don't know what 6 is. Rules other than Conway Life also have 3 and 5, and possibly other states still that I haven't seen.

The puffer notation issue is a tough one. The minus sign is a consequence of the method used to determine uniqueness. I don't think the puffer notation is very illuminating, at least all I understand is yl[period of growth]. The following two values don't make sense, and then I think what follows is an apgcode but I'm not sure. I don't believe there are plans to change it, but it could be nice to make it more understandable. Plus it is concerning that the protocol wasn't quite robust enough to handle a rogue glider in the GSPE + debris pattern, and also that it labelled the zz_LINEAR as it did.

The second number in a yl code is another period, and the third number is a "moment", as returned by deepperiod(). In fact deepperiod() returns several moments, but for reasons I do not understand the second and third are not given verbatim; instead they're joined with a hash sign, and the MD5 of the resulting string is then used.

[gameoflifeboy]

Well, there were a few more interesting things about Catagolue that I didn't post, but now that there's a thread for these, I guess I can post them here.

• There is a way to run any pattern for up to 100 generations. Just replace "100" in the URL with the number of generations you want, "b3s23" with the rule, and "172" with an encoding of the pattern. If it is a spaceship that repeats itself in less than or equal to the given number of generations, it will render as a spaceship without having to enter xq. This gives some hints as to how apgcodes are parsed into SVGs.
  WARNING: The SVG seems to run best on Chrome.
• As a matter of fact, there are pages for objects be bigger than 40 by 40: http://catagolue.appspot.com/object/xp2_rhewe44ewe44ewe44ewe44ewe44ewe44ehr/b3s23 In fact, I once saw Calcyman mention that he was thinking of increasing the maximum bounding box of parsed objects in apgsearch.
• Not entering a rulestring in /hashsoup after the soup seed prints a rule-neutral version of the soup, with periods as off cells and asterisks as on cells. Back in early Catagolue history, this was the way the sample soups were linked from the object page: http://catagolue.appspot.com/hashsoup/D8_4/m_7tKhznGbZpTi4539278
• Some comments can break the page.
• All different "symmetries" have different colored sample soups, even those that were errors. This suggests that the color is made by taking a function on the name of the symmetry, except for C1 which gets black. Here is the list of symmetries and the colors their soups get:
  
  Code: Select all

  C1 #000000
D4 +4 #d32f3f
25pct: #a8b75d
8x32: #6d0ecf
C2_1: #f83e05
C2_2: #31a6d8
C2_4: #aceb02
D2_+1: #39bab9
D2_+2: #747d16
D2_x: #fb71fe
C4_1: #d085ff
C4_4: #cd14a0
D4_+1 #f6b2b6
D4_+2: #f8e612
D4_+4: #cfc20e
D4_x1: #ae360f
D4_x4: #3e5b59
D8_1: #ed65b6
D8_4: #a621fb


Minus-containing puffer: http://catagolue.appspot.com/object/yl1 ... 3dd/b38s23

I pointed this out on page 13 of the apgsearch v1.0 thread; did you get this from there?

[drc]

It's also possible to nest HTML tags,
Also, putting an iframe tag can disable comments.
You should only test this on obscure object pages though,

[blah]

There's no data sanitation in the comments section. Fun.
Edit: just to clarify, I did that.

[Apple Bottom]

There's no data sanitation in the comments section. Fun.

Ouch, that's bad. Have you let Calcyman know? The above is just harmless fun (mostly), but this definitely needs to be fixed.

[calcyman]

Why is this even happening? I expected pegdown would by default sanitise any instances of < and > to gt and lt:

https://github.com/sirthias/pegdown/

I think I need to use the 'INLINE HTML SUPPRESSION' and 'HTML BLOCK SUPPRESSION' flags.

[calcyman]

Why is this even happening? I expected pegdown would by default sanitise any instances of < and > to gt and lt:

https://github.com/sirthias/pegdown/

I think I need to use the 'INLINE HTML SUPPRESSION' and 'HTML BLOCK SUPPRESSION' flags.

Fixed.

[blah]

Fixed.

Kind of. It actually seems to remove html tags, rather than sanitise them. Like, it doesn't even say "<h1>blah</h1>", it just says "blah". You can't see what my original comments on that page were.

Maybe that's good enough for you, I guess that's all that matters.

[Apple Bottom]

Another oddity...

b36is2-i3eiq4ay has a "C1-2016-04-05-2016-04-05" symmetry. This one looks like a census backup, and in fact displays a message to this effect, but it is listed on the rule's list of symmetries. What this ultimately means only Calcyman knows.

In the same vein, b368s245 has a D4 symmetry. What kind of D4, I hear you ask? Judging by a sample sample soup it's actually C1 in disguise.

[GUYTU6J]

Nobody noticed those ads in this page?

[Saka]

xq0 and yl0 exist, but xs0 does not.

[Apple Bottom]

xq0 and yl0 exist, but xs0 does not.

Where do those exist?

[Saka]

xq0 and yl0 exist, but xs0 does not.

Where do those exist?

Put them in the objects page with any random string after the _
(e.g. xq0_foobar)

[Rhombic]

This one is rather worrying.
http://catagolue.appspot.com/census/b2- ... s2a3-i4/C1

If you check the xp2 list, xp2_7 and xp2_5 are listed separately in spite of being the two phases of the same oscillator.

[A for awesome]

This one is rather worrying.
http://catagolue.appspot.com/census/b2- ... s2a3-i4/C1

If you check the xp2 list, xp2_7 and xp2_5 are listed separately in spite of being the two phases of the same oscillator.

Yes, I know. This is due to the assumption of the ExpungeObjects rule that the blinker functions normally, which it does in all outer-totalistic rules. I will hopefully be able to fix this (and other issues) soon, as I will finally have a computer that is able to run Golly again (hopefully) by next month.

[drc]

Two new ~20x20 c/3 and 2c/5 spaceships have occured...30000 times in total...in a symmetry called "Saka_Test"

EDIT:

Code: Select all

On 2016-12-18 at 13:10:25 UTC, Someone wrote:

Sorry guys (This is a test with a hacked apgsearch)

On 2016-12-18 at 13:40:19 UTC, Someone wrote:

Sorry

Welp. Still a bit of an oddity I guess.

[Saka]

Two new ~20x20 c/3 and 2c/5 spaceships have occured...30000 times in total...in a symmetry called "Saka_Test"

EDIT:

Code: Select all

On 2016-12-18 at 13:10:25 UTC, Someone wrote:

Sorry guys (This is a test with a hacked apgsearch)

On 2016-12-18 at 13:40:19 UTC, Someone wrote:

Sorry

Welp. Still a bit of an oddity I guess.

[yootaa]

What's this?
https://catagolue.appspot.com/object/xs123_g8861v0ccz11dd1dkjhe0ehjkczwcc0cai2s0s2iac0ccz355p1v0ddxdd0v1p553zy011y511/b3s23

[Saka]

What's this?
https://catagolue.appspot.com/object/xs123_g8861v0ccz11dd1dkjhe0ehjkczwcc0cai2s0s2iac0ccz355p1v0ddxdd0v1p553zy011y511/b3s23

Hmm yes it's missing the top right quarter