#
Date: Tue, 22 Mar 2005 18:45:52 +0000 (GMT)
Subject: Techinfo woes

Wow, I've just spend a day fighting with one of the *nastiest*, most
ill-thought-out, deliberately obfuscated pieces of horrendous web
structure that I have seen in a *long* time.  I'm talking about
techinfo.toyota.com, where I just did the one-day subscription long
enough to web-whack the full manual sets for several vehicles.  I
managed to semi-automate most of the process once I was able to obtain
the indexes, but migod what a painful experience in general.  Is there
anyone from Toyota on this forum that has anything to do with their
totally crocked notion of web design, or can pass things on to the
right people?  This *really* needs to be 100% ripped out and redone.

First off, relying on browser scripting is NOT w3 standards-based, and
forcing the user to totally drop his drawers from a security standpoint
is bad practice in general.  Especially when as it turns out the tree
of PDF files can easily be indexed into a few ordinary HTML files,
instead of trying to force everything through pieces of javascript
triggered by an onLoad="..." hook.  In many cases, that BREAKS
functionality and causes the browser to never find the .PDF at all.
Lose the .pl wrapper rot already, it doesn't gain anyone anything
and wastes that much more bandwidth with an additional request.

Two or three separate browsers on two different OSes returned differently
broken results when trying to parse the "leftnav" stuff after selecting
vehicles, years, and manual section types.  The most amusing failures were
the "access denied" errors from it trying to open the ...whatever/leftnav/
directory itself.  Finally making it work well enough under a different
browser allowed saving the "search results" frame [again, why bother
presenting it that way, especially if you can't make the scrollbars
consistently work right?] into a file, and that was enough to convert
into the actual PDF filenames to plug into "wget" and start suckin'
them down. In case anyone's curious, the transform is to convert
something of the form

  cgi-bin/ns-data-display_pl/,0/r1/ileaf/DIRNAME/rest/of/it/FILE.pl

into

  ileaf/DIRNAME/DIRNApdf/rest/of/it/FILE.pdf

and that's !finally! the PDF in question.  The additional directory
element that must be added is obtained by repeating the first five chars
of the DIRNAME and adding "pdf" to that.  Given the "search results"
HTML output, it's relatively easy to convert into 1> a list of plain
URLs to retrieve the PDFs, and 2> a nice straightforward HTML index
into the ileaf/ tree telling what each file is.  I did this for all
the resultant sections, such as "service bulletins", "repair manual",
etc and it's *so* much nicer and faster to wander through locally with
a browser and read the stuff.  Yay "sed".

And where is this vaunted TIS "keyword search" facility?  I never saw
where to access it the entire time, meaning I had to ask for one
manual section at a time per vehicle per year to return anything
like a useful index.  I guess I'll have to generate that for myself
by ripping the text out of all the PDFs and doing searches, huh?

Hellooooo Toyota, please take this seriously and LEARN from it.  All the
world is NOT microsoft and internet explorer, in case you haven't heard.
You could save both yourselves and your customers an awful lot of trouble
by greatly simplifying this stuff.  I should be able to use a TIS
subscription with a simple, fast text-based browser, and if you're not
testing functionality with multiple clients then you're not anywhere
close to industry best practice.  I'm not looking for a bloody "web
experience" as your designers may have the misguided notion that I might
be, I'm looking for INFORMATION.  As is anyone else who comes to your
site. Even your dealers freely admit that the way you present this is
"a little weird", to put it as charitably as possible.  Well, having seen
what I as an end user can reduce this to in stark contrast to the mess
you've made of it, I will put it even more simply: IT SUCKS.  I won't
begrudge you the ten bucks since I did finally manage to get information
out of it, but frankly, you ought to be paying ME for the time and high
frustration I had to work around to do it.  Go find the web-server
logs from my activity yesterday, starting around 3pm Eastern, to use as
a guide as to things that can go wrong -- it shouldn't be hard to find
the username and/or IP address in question.

It is very tempting to offer the reduced trees on CD, or pointers to
compressed tarballs thereof, to anyone who can prove to me that they've
bought a Techinfo subscription, if I happen to have the vehicles of
interest or something close enough, just to save them the pain of having
to actually use the site as it stands.  What would your lawyers' take
on that be?

_H*