Colourful friends

[lnr]

acronym addedentry aendr ailbhe aldabra ali_anarres anastamosis andyroyal angoel angua antinomy arkady arnhem atreic barberio barnacle beckyc beingjdc bellinghman bellinghwoman bigtonyd bjh21 blackherring bootpunk bopeepsheep brrm burkesworks catyak ceb chickenfeet2003 cjwatson claroscuro crag_du crazyscot damerell daneel_olivaw davefish deadpidge deliberateblank dinesha dmwcarol dreamingchristi drswirly echoing_void ejde elise emarkienna emperor ewx exotic_ices eyekiller fanf feanelwa fivemack florence90 fluffymark fluffymormegil games_puzzles geekette8 ghoti glamwhorebunni glitterboy1 gnimmel hampsterboy hoiho hollyp hsenag imc isihac j4 jbmurray jiggery_pokery juggzy julietk jvvw k425 k425sbug kaet kaet_dreams karen2205 kathypuss keirf keris kitty_goth lark_ascending lethargic_man liarlyre littlemissgoth lnr lusercop marnameow marnanel martling marypegg meirion mhk mikewd mjg59 mobbsy mpinna mr_tom mst3kgirl mstevens mtbc100 naath naranek nassus natural20 newcat nilasae nja northernmonkey nou nslm nwc10 oldbloke other_faces p_a_r_a perdita_fysh pir pjc50 pm215 pookee pseudomonas ptc24 pto452 q_skud_ razornet redcountess reddragdiva rejs rillaith rmc28 robinbloke rysmiel sampiano saraphale sbp scat0324 senji shadow_jess shereenb simonb simont sion_a sirpropaganda skorpionuk smallclanger songster sphyg sweh taimatsu tamsinj teleute the_mendicant thorfinn tienelle timeplease tlingel truecatachresis uitlander uon vatine venta vyvyan wintrmute womble2 yonmei zenithed cantabrigiensis kittenherder livechiark paidmembers pie_scoffers slimmingworld uk_misc acroblog dotaturls nsfw stephdiary

NB: this is done by replacing each name with <font color=name>name</font>

Comments

[simont]

Ah, I'm glad you did that. I was starting to wonder whether all of this had gone straight over my browser's head, because both you and senji looked profoundly uninteresting under this transformation. Now I can actually see most of the colours.

[mobbsy.livejournal.com]

IE seems to be a bit more keen at colouring things than Mozilla, however both lnr and senji appear black there too.

[mobbsy.livejournal.com]

While Konqueror appears to just not show colours for random names.

[lnr]

Indeed, that's exactly why I did it :-)

[simont]

:-)

So has anyone worked out what the algorithm is yet? (Looking in a browser's source is cheating.)

The fact that practically everything in the A-F range is coloured and many things beyond there are black suggests to me that it's something to do with extracting valid hex digits from the supplied string; and indeed on closer examination with xmag, "senji" is a slightly reddish #0e0000 rather than genuine black, so the "e" has clearly had an effect.

However, it doesn't seem to be just a matter of extracting the first six valid hex digits; a good example is "claroscuro", in which the "c" has had an effect in forming the colour #c00000, but the "a" two characters later has done nothing. Curiouser and curiouser.

[rillaith.livejournal.com]

I could be wrong, but i believe it does something along the line of translate it into hex and then back into a colour. Probably via some mayo and mustard just to confuse it a bit more. And don't forget the duck.

[mobbsy.livejournal.com]

Now I'm hungry.

[lnr]

Quack!

[imc.livejournal.com]

Before I read these comments I wondered the same thing. I'm currently using Mozilla in the form of Netscape 7 for Solaris. I should think this is heavily browser dependent, of course. Browsers that refuse to colour most of the names are probably acting most correctly. The browser probably shouldn't be trying to extract valid hex digits unless the name starts with '#'.

There's a lower limit to the number of characters which Mozilla will recognise as a colour. I believe that limit is 4; shorter names all seem to be black. The basic algorithm seems to be to translate each invalid letter to zero and pad on the right with zeros, hence becomes #0ae000 and becomes #b00210. However, if the name is 7-9 characters long, Mozilla skips every third character (which would be the least-significant nybble of each channel in a 36-bit colour specification), so becomes cr[a]zy[s]co[t] which is #c000c0. With longer words, it skips two characters, so becomes ad[de]de[nt]ry which is #adde00.

[bopeepsheep.livejournal.com]

That probably explains the way the bisonfamily colours came out (see my (http://www.livejournal.com/users/bopeepsheep/) list). I clearly don't know enough about hex, because I'm wondering why names at the beginning of the alphabet are all orange/red and those at the end blue/green. And why does it work when you look in the main view but not when you look at a specific entry?

[sparrowsion]

It's working in specific entries for me (Opera).

Hex values for colours (in this context) consist of three numbers: the amount of red, green and blue, each number being two hex digits, so value for each ranges from 00 (none) to FF (maximum). Anyone with a name starting in the range a-f is therefore going to get a colour with a large red component (at least A0). Whereas anyone with a name starting after that is going to have a smaller red component (at most 0F). Amounts of green and blue in the two sets of names are probably about as likely -- but whereas in the first set they tend to shade the colours towards yellow and purple, in the second set they provide all the colour information, so you see greens, blues and blacks.

[imc.livejournal.com]

It looks like Mozilla switches to a more strict colour parsing if the document has a valid DOCTYPE declaration at the top of it. [Actually, that's not quite true - Google found this description (http://www.mozilla.org/docs/web-developer/quirks/doctypes.html) of Mozilla's DOCTYPE sniffing.]

In LJ's XColibur style, the `specific entry' view has an XHTML 1.0 Transitional DOCTYPE, while the `recent entries' view doesn't have one. Hence the difference in appearance.

[bopeepsheep.livejournal.com]

Thank you, I actually understood that explanation. :-)

I had guessed there was some logic to it but hadn't a clue what it was, since I could see plenty of as (etc) in later names...

[rillaith.livejournal.com]

Very possibly. Plus some browsers have a hefty list of text names that they will interpret (http://www.w3schools.com/css/css_colornames.asp)...

[jiggery-pokery.livejournal.com]

Ooh, champion!

Hence jiggery_pokery becomes JI[gg]ER[y_]PO[kery] or #00E000, the rather fetching shade of green, in Mozilla Firefox. In MSIE, it's a very dull shade of blue. I particularly like karen2205 becoming KA[r]EN[2]20[5] and hence #0AE020. That's pretty nifty. (Come to think of it, the date is probably a birthday, which means it must be Karen's birthday soon. Must not forget between now and then.)

Thanks for the explanation. This is why people are saying "ooh, look at that, I'm a lovely shade of blue!" and my response is "No... that's... red." MSIE, eh?

a dull shade of blue

[imc.livejournal.com]

I had a brief look at MSIE via a remote desktop and it seemed to agree with Mozilla on most of the names, but the longer ones differed and I didn't discern much of a pattern (though the colours seemed to be made up of the digits `0' and `c').

[marnanel.livejournal.com]

http://bugzilla.mozilla.org/show_bug.cgi?id=121738 discusses this. Not sure how up to date it is.