Comments on: Whole-text DOM functionality and Acid3 redux

By: James John Malcolm

James John Malcolm — Tue, 23 Mar 2010 12:05:06 +0000

“curser:none” may itself become a workaround (or part of an ingenious solution) for another problem…that’s my point.

By: Sobre Internet Explorer 9, estándares y pruebas varias | MuyWindows

Sobre Internet Explorer 9, estándares y pruebas varias | MuyWindows — Tue, 23 Mar 2010 00:25:08 +0000

[...] sistemática. Es decir, no refleja el uso real de un navegador. Es interesante leer al respecto una entrada del blog de Jeff Walden (de Mozilla) en el que confiesa que se agregaron partes al motor de render de Firefox simplemente [...]

By: Jeff

Jeff — Mon, 22 Mar 2010 20:31:56 +0000

I wasn’t intending to communicate such, although a few times I did use less common words when I had the opportunity to use them, for the sheer joy of doing so. Buckley’s closing argument when discussing his use of intelligent vocabulary resonates with me. Not every argument must always use the plainest words possible, particularly not when the stakes are low as here. If I fail to convince a reader here due to somewhat-esoteric word choice, the worst of the consequences is that someone disagrees with me. For this argument, I can live with that.

Still, I’m happy to find that, on reread, you think it may have just been a first-time mis-impression. If you do happen to see anything else in the future, feel free to say. It should be possible to flex a vocabulary without putting off too many people if one takes a little care.

By: Jeff

Jeff — Mon, 22 Mar 2010 20:20:38 +0000

I thought it was spam on first read, but as it didn’t link anywhere, and as it included an email address that looked like a personal email address (first initial, last name, that sort of formulation), I decided otherwise at the end.

I think in many cases you can know when an edge case can and can’t be used as a terrific solution: if the edge case can be duplicated another way with little effort (transparent cursor), if the process for invoking it is baroque (num.toString(undefined)), at the least, it’s not a terrific solution. It may be marginally better for aesthetic reasons, but aesthetics don’t put bread on the table like something new, properly implemented, can.

By: namename

namename — Mon, 22 Mar 2010 18:01:26 +0000

For some people their ego is like a visible cloud following them around, all their actions and thoughts and words are self indulgent and self-affirming.

I meant your writing has this feel of trying very hard to communicate how smart you are, in addition to the original actual content or intent (Acid-DOM-etc).

I just skimmed through it again though, I’m probably wrong. Just ignore these sniping petty comments.

By: James John Malcolm

James John Malcolm — Mon, 22 Mar 2010 09:11:36 +0000

Re: Namename’s comment: I think it’s a spam comment.

Re: Acid:
Seems like we generally want the same thing, although I still think there are cases where it’s valuable to implement several sub tests to test a number of corner cases to make sure that browsers implement a feature completely and properly.

The main crux of my argument for this is that you never know when an apparent edge case can be used as a terrific solution.

Imagine if margin(-right/left):auto hadn’t been implemented interoperably (which isn’t a word, but you know what I mean)!

By: Jeff

Jeff — Fri, 19 Mar 2010 10:41:19 +0000

Thanks for the compliment. I put more effort (more importantly, better effort) into polishing this post than I have into others; I think it has specific areas where quality is quantifiably better than in past posts.

I’m curious why specifically you suggest “big ego”. Aside from it being my own opinion and actions and therefore being an expression of egoism (or perhaps egotism? I find the precise distinction elusory), I’m not sure how it is notably egocentric.

By: Jeff

Jeff — Fri, 19 Mar 2010 10:25:06 +0000

I didn’t mean to suggest subtests in high-profile Acid tests should be non-specific: rather, only that the general target of each subtest should be something where browsers disagree widely (thus making the feature impossible to use) due to mis-implementation or non-implementation. When writing the precise subtests it’s perfectly fine (great, even) to go for the corner cases, edge cases, and inter-feature interaction points. So, for example, rather than just test that generated content “shows up”, you test its layout in interaction with nearby floating content, or with column breaks, and so on. You should still hit the general case while you’re at it — but try not to spend more than one situation in the group of subtests with that as the primary goal.

Yes, one missing hard corner case is annoying. However, I would rather have, say, new and well-tested CSS media query support (Acid3 did an excellent job on this point) than mindless testing that the laundry list of cursor keywords all work, once the threshold of arbitrary custom cursors is supported.

I’m willing to grant Acid2 leeway on some trivialities because in broad strokes what it tested was desirable and not implemented (or was not interoperably implemented).

By: James John Malcolm

James John Malcolm — Fri, 19 Mar 2010 09:46:04 +0000

That’s what I meant to write, Acid 4. Acid 3′s been done for a long time.

Why waste effort on something developers could do in a slightly different way with very little extra effort?

Because stuff should work properly and predictably. When it doesn’t it’s a pain in the ass for designers.

I don’t think high-profile tests should only test for new, not easily replicated, functionality in a non-specific way. New non-complete features are being added to engines all the time anyway, without tests driving them. Having a high-profile test make sure things really work the way they should is very valuable in my opinion.

Acid 2 btw, included many parsing ‘trivialities’ like /* comments with backslashes \*/.

By: namename

namename — Fri, 19 Mar 2010 09:25:16 +0000

Smart and good writing, albeit big ego.