Sent by aardvark on 28 November 2002 08:08
> From: Adam Kuehn [EMAIL-REMOVED]>
> >now *that's* new, and also sounds more like an excuse to pardon
> >yourself from a rule that can sometimes be inconvenient...
yeah, that sounded more caustic than intended...
> A properly designed website should have a solid structural foundation
> or your users will be hopelessly lost. That structure can provide a
that's IA... things like human-readable folders (and hence, URLs),
logical breakdown of document locations, etc...
> built-in context for individual elements within the structure, whether
> those elements are pages or elements within pages. Well-designed pages
and that's the HTML...
> also provide a structure of their own that informs the use of the
> elements within them, but that structure is a secondary structure and
> in rare cases must be subordinate to the overlying site structure.
> The site structure is the main framework you are building. That seems
> axiomatic to me.
and it seems self-evident to me that HTML is a page/document
level markup language (hence the notion of a Document Type
Declaration)... yes, i said that before, but you didn't address that
tidbit at all, and i feel it stands at the heart of proper use of the
elements... ignoring that fact leaves all sorts of room for its
does this mean if i have a quote as a pop-up that i can encase it in
<blockquote> and leave out the <html></html> tags? it is, after
all, simply an extension of the parent page, and the site is, after
all, one contiguous document...
that's the thing, this logic sounds too convenient to me, and too
open to abuse... that's why i can't embrace it...
> Furthermore, very long documents are frequently split across pages for
> ease of use. *Usually* the pages can split so as to correspond with
> header levels. Sometimes, however, for particularly long and
> complicated documents, this doesn't work well. So I split at a
> sensible division point within the sub-section.
two problems with that notion for me...
1) my own user testing has shown that readers *don't* want pages
split across screens... it makes it harder to print, scan visually,
and requires more URLs... the main reason bigger sites do it is for
more ad impressions (and revenue)... after converting a lot of my
articles back to one-pagers, i decided i'd toss that approach out
the window... not to mention, every time that topic comes up
elsewhere, i generally (always) hear people against split articles...
2) *if* i were to do that, i would consider a different usability
perspective... if iu've just split the page deeper in a section, how do
i know my reader will maintain context on to the next screen? how
do i know someone coming to that URL from an SE/email will know
where the heck they are? simple, i'd recap:
<h1>My CSS Article</h1>
<h2>Shorthand Properties, continued</h2>
now *that*, IMO, is the right way to do it, from a usability
perspective, if you have to split... coincidentally, that also
eliminates the issue of skipping levels...
> If that sounds like an excuse to you, that's OK with me.
i know that's a valid response to my earlier statement, but i do
stand by my belief that it's an easy way out of applying proper
hierarchy to a page...
> >sure, if the site as a whole were part of the spec, then maybe, but
> >the X/HTML specs are for applying semantic/structural mark-up to the
> >page, which is the individual document... i can't reconcile your
> >argument with that fact, so you're going to have to help me out...
> Why would adding the site level to the specs change your entire way of
> thinking about site and page design? The specs are supposed to inform
it wouldn't, it would change my way of applying the specs... the
"site level" is independent of the markup on a page, it's simply the
location of a file... good IA makes it a seamless fit, not code...
> your design, not substitute for it. Besides which, even if you are a
> slave to the spec, the spec doesn't disallow the practice of skipping
> heading levels, whether you are writing HTML or XHTML. So if *your*
> argument is based entirely on what the specs say, you haven't exactly
> justified your position. My "excuse" is pardoning me from a "rule"
> that isn't even an official rule.
no, i never said skipping levels was verboten... i don't feel it's
appropriate, and is almost always an easy way out of good
structure or is a misunderstanding of the structure of the page (or
even a poorly-structured page)...
> I could do that, but it doesn't look correct to me, perhaps because of
> the practice I see applied in the print world. An article in a
you can understand that everything you say after that just got
filtered by my brain, i hope... as soon as i saw "print," i've fallen
under the (hopefully false) assumption that you are trying to treat
the web as print...
> magazine, for instance, may repeat the article name in a small header
> across the page top, but secondary headers are virtually *never*
> repeated, even if a secondary header spans a page turn. Of course,
a magazine is not often read a page at a time by a user who picks
the pages off the floor... a search engine has that effect... you can
come into any page out of context... just because your site
contains them all doesn't mean they are bound in a linear, stapled
> the web is more flexible than the printed page, so you have some
> flexibility in choosing exactly where your page breaks are. But
> sometimes the sub-sections are just too long to rest on individual
> pages and require splitting. If that's the case, I split the article
i'm still struggling with the assertion that the pages are too long...
i've tested this theory many times over, and users have always
preferred one-page articles over clicking "next" every couple
> as best I can and dive right in with an h3. I suppose I could repeat
> the h2 and use display: none. That solution never really occurred to
> me before this discussion.
that's an interesting one... kind of a neat solution... sounds
hackish, so i doubt i'd use it, but that's still neat...
> Of course, the real problem is that a lot of articles are not written
> expressly for the web, whether because authors are conditioned to web
> writing, or linear, non-condensed writing is somehow more natural for
> people, or whatever. I've never run across an article purpose-written
> for the medium that requires non-consecutive headers. But I digress.
> >perhaps your page split isn't semantically correct if you're starting
> >with <h3>s... it means you're splitting the page too soon, right in
> >the middle of a section...
> Or perhaps I have pretty good, user-tested reasons.
still stuck on that...
if you do have any links to tests on that, or can relate your own
user testing, i'd love to hear it... i'm always looking for results that
differ from my own to explore methodologies and the like...
alright, back to code...
Read the evolt.org case study
Usability: The Site Speaks for Itself
Sponsored by www.westciv.com - CSS resources | software | learning