[Wikipedia-l] generic markup and miscoded templates

[Brion VIBBER]

lcrocker at nupedia.com wrote:
>>Level-three headers (===) are well established as the standard
>>header on Wikipedia. If you hate this so much, better to simply
>>redefine it to pump out ideologically correct H2 tags instead of
>>H3 rather than to prescribe the change of thousands of pages and
>>demand a change in markup behavior.
> 
> I don't think it's "well established" at all.  I always use
> == for my first subheads, and many others do as well.

Well, let's take a look:

SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]== [^=]+ ==[^=]'
Articles that use ==: 1430

SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]=== [^=]+ ===[^=]'
Articles that use ===: 4260

SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]==== [^=]+ ====[^=]'
Articles that use ====: 1008

SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]===== [^=]+ =====[^=]'
Articles that use =====: 1

SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]====== [^=]+ ======[^=]'
Articles that use ======: 1


SELECT COUNT(*) FROM cur WHERE cur_text REGEXP '[^=]== [^=]+ ==[^=]' and 
cur_text REGEXP '[^=]=== [^=]+ ===[^=]'
Articles that use both == and ===: 376

SELECT COUNT(*) FROM cur WHERE (NOT (cur_text REGEXP '[^=]== [^=]+ 
==[^=]')) and cur_text REGEXP '[^=]=== [^=]+ ===[^=]'
Articles that use ===, but not ==: 3883

So there are more than twice as many pages using === as there are using 
==, and more than *ten times as many* pages using === but not == as 
there are using both == and === (and that would thus need to be 
changed). That's what I call "well established".

>  There's
> a nice simple correspondence between ==/H2, ===/H3, etc.  H1
> is reserved for the article title.  If we shifted === to 
> produce H2 as you suggest, then what whould == produce?

We already discarded the = header entirely (which used to work in Usemod 
days) to little weeping, requiring anyone who might have used it to 
change to == or ===. The exact correspondence between level numbers 
isn't particularly relevant as long as they're supposed to be hierarchical.

> I have heard complaints that people think H2 is rendered too
> large; that can be fixed with stylesheet changes, and I'm
> certainly open to doing that.

Certainly.

 > But let's settle on a standard
> for reasons other than mere inertia.

Got something against inertia? That Wikipedia has it is good; it keeps 
the project going. :)

-- brion vibber (brion @ pobox.com)