hi 

i got a few mails from people who checked the rc1. a few of the issues we
knew already. please find enclosed a digested feeback text file which at
least discusses some issues i think we should/could look at when the next
version rolls out (:-)). really they are not ISSUES, but more NICETOHAVES,
maybe we can have such a file at perl.apache.org/NICETOHAVES ?

./allan

[Tom snipped]

Those are among the most attractive web pages I've ever seen. Nice work.

I have noted (and appreciate) the rules which guided your work. I subscribe to 
the exact same rules, as it happens.

As far as fonts, I question the rationale in specifying a size of 1em for the 
body. Wouldn't browsers assume that? And personally, I wouldn't suggest named 
fonts at all. My Helvetica, for example, is a PostScript version which renders 
web pages very poorly. I would prefer that my browser's default sans-serif font 
be used instead.

Many of your visitors will be using Lynx, I expect. (I, for one, am always 
using Lynx when I am configuring a server and need to consult the 
documentation.) You might want to direct some effort into making the pages look 
a little prettier for them. 

[end Tom snipped]


************************


[Terje double snipped]

That's ok. If you are getting ready for release, now is a
good time to start tentative planning for the next major overhaul. I'll
make my comments in that light.

1) Please, *please*, reconsider the design decision that put the 
navigation(-ish) boxes on the left side of the page!

> by navigation(-ish) boxes, do you mean the prev|up|next currently on >the 
> righthand side of the screen?

I mean the "Slashbox"-style boxes down the left edge of the page. They simply 
eat up too much of the valuable left margin of the page. The margin in 
print/layout works because it's _empty_ space; it creates some air on the left 
of the main body text. When you put "stuff" there you create a distraction 
which intereferes when you try to read the main body text. Placing those boxes 
on the right side takes care of this problem as well as degrading more 
gracefully when printed or viewed on low-resolution screens.

> we want to produce a seperate print style sheet, that hopefully will take 
> care of what you describe above

Separate stylesheets for different media are a good idea, yes. I would prefer 
to move those boxes for on-screen presentation too though, for the reasons 
above. If you feel very strongly about keeping those boxes where they are, you 
may consider providing an alternate stylesheet that puts the boxes on the right 
(Mozilla-based browsers lets you choose which of several alternate stylesheets 
to use from a menu).



2) Disregard Netscape 4.x *completely* as it'll only give you false ideas of 
what "works" and what doesn't. Hide *all* CSS from it with <style>@import 
foo.css</style> or some such trick and make sure only the _content_ gets 
through even if it's ugly as all hell.


3) State explicitly somewhere that it is _acceptable_ if the _presentation_ is 
less then perfect in non-conforming browsers so long as the _content_ is always 
available.

It will save a lot of pain later if you explicitly state, as part of your 
design goals and criteria, that it's _ok_ to have some small display glitches 
in less then perfect CSS implementations as long as the content comes through 
allright.

Graphic designers tend to scream bloody murder at such glitches (because they 
really are bloody ugly, and badly ruins the aestetics), but the big mistake is 
to then try to fix them with wide-reaching workarounds that lowers the overall 
technical quality of the code. If it's been decided before-hand that these are 
acceptable, people tend to have a more realistic attitude to what constitutes 
acceptable tradeoffs between technical purity and practical hackery.

For instance, your choice of the Transitional DOCTYPE is probably due to a 
percieved need for "Quirks" mode in browsers to work around poor standards 
support in other browsers. I've found that by setting the goals realistically, 
I can dispense with such hackery and use a Strict DOCTYPE and still achieve 
good results.

I'd point to <URL:http://validator.w3.org:8001/> as an example, but I fear I'm 
too little skilled as a graphic designer to be able to impress any such with my 
work. :-)

The site should however be sufficient to indicate that it _is_ possible to
find a marriage of the two opposites that satisfies both if only you find the 
right balance in your compromises. A skilled graphic designer would have been 
able to improve those aspects of the site rather much and still allow me my 
fanatical adherence to standards and good practices.


4) Perform a WCAG review of the site! This *will* show you major problems. I 
can almost guarantee you that you will want to
stick your head in the sand, mutter "nahnahnahnahICan'tHearYou!nahnahnah", and 
pretend they do not exist. Please don't. Accesibility is
important and the only reason you're in trouble now is because you didn't take 
this into account from the start.

Try <URL:http://valet.webthing.com/access/> for example.

The form defaults to only performing Normalization of the code. You need to 
select a test suite before it will give you any results. Try checking your 
pages with "WCAG-AAA" selected. It will turn up a *lot* of issues. Fixing some 
of them will be a pain, and some can't be fixed without major restructuring on 
your part. Don't let that discourage you, however. Start off by picking the 
low-hanging fruit; as a little accessability is better then no attention to 
accessibility. Once you get started it's also possible to ask for volunteers to 
do a review and help you with improving you accessability. Give me a holler and 
I'll give you a few pointers; or work your way from 
<URL:http://www.w3.org/WAI/>.


5) You use a "<link>" element to refer to the "bookmark icon", but ignore the 
navigation ("top", "next", "previous", "toc", etc.)
<link>s that lets Mozilla and other modern browsers provide the navigation.

> i dont get you, exactly what is wrong with that ?

<link rev="start" href="/">
<link rev="next" href="chapther2.html">
<link rev="glossary" href="glossary.html">

These let you define logical relationships between the various pages on the 
site -- especially for logical collections of pages such as
the manual -- that will allow smart browsers to provide an UI for moving 
between those parts. Read up on the Mozilla "Links Toolbar", or the similar 
feature in iCab. Search engines can also take advantage of this sort of markup.

You may also want to look into using "Dublin Core Metadata" to add information 
about modification times, author, copyright, etc. etc. to the pages. This is 
also good for automated processing -- be it by search engines or by browsers -- 
of the pages.

IOW, this isn't really about the "navigation links" as they appear in the page 
as shown on screen. This is about structural metadata in the <head> section of 
the document.

Please don't think that I'm slagging your design! It's just that it's very hard 
to "critique" something without also "critisizing" it. Hopefully I've managed 
to avoid the worst pitfalls and gotten through that I'm suggesting ways to be 
"even better", not trashing you for not doing better in the first place (which 
would imply that your work wasn't very good). As I wrote, the work you've done 
_is_ very good. What I'm "critiquing" isn't so much the work as the decisions 
made when you hammered out your goals and success criteria up front.


[end Terje double snipped]

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to