a UI rant

[cellio]

(I posted this rant on the company wiki, on the aptly-named "rants" page, but I'm going to share it with a wider audience.)

HTML has been in common use for more than a decade. The field of UI design has been around for several more. Surely, somewhere in there, most people got the clue that when displaying text, you specify both or neither of text color and background color (with strong arguments for "neither" to give the user some control).

I was a little surprised to find that Sun does not have this clue, until I switched my environment to a reverse-video scheme and then looked at some Javadoc. Tan text on white background -- goody! -- because the HTML sets BGCOLOR=white and is silent on text color. But wait, it gets better -- they also do it for table cells and rows! Now I have to maintain a local style sheet with these three changes, and re-copy it into the output directory every time I geenrate Javadoc, because Sun decided to set half of this pair while fetching the other half from the OS.

There's no excuse for anyone to be making this egregious error in 2006.

Comments

[astroprisoner.livejournal.com]

That is astoundingly stupid of them.

[cellio]

Yeah.

I realized after posting this that I can make it less annoying by writing a new ant target that wraps the javadoc build and then copies the stylesheet (so I can get local builds that I can read without screwing up the official build). But darn it, I shouldn't have to...

I wonder how easy this is to fix in a custom doclet. (We already have a custom doclet, but I haven't studied the code lately.)

[http://users.livejournal.com/merle_/]

I have never looked into it, but I would assume that whatever process creates the JavaDoc pages works off of a set of templated fragments -- in which case it could be as easy as finding and modifying the fragments. That technique has worked wonderfully for me in many applications so far. The catch being, of course, that you have to find those fragments...