- From: Ian Hickson <ian@hixie.ch>
- Date: Thu, 7 Jun 2012 20:31:59 +0000 (UTC)
- To: Henri Sivonen <hsivonen@iki.fi>
- cc: "Tab Atkins Jr." <jackalmage@gmail.com>, Jonas Sicking <jonas@sicking.cc>, Rafael Weinstein <rafaelw@google.com>, Yehuda Katz <wycats@gmail.com>, Webapps WG <public-webapps@w3.org>, Scott González <scott.gonzalez@gmail.com>
On Thu, 7 Jun 2012, Henri Sivonen wrote: > > I believe in this case not changing the way SVG script content tokenizes > would be best for authors. For what it's worth, I agree with Henri here. In my experience, spec churn is the number two way of making a spec fail. I think it's better to have something that works consistently everywhere than to have things work different across different browsers and even different versions of the same browser. That's the effect of spec churn. It also has the effect of putting test suites in unclear states, which is especially bad for test suites that have been copied into browser vendors' development environments (especially if they don't realise the spec has changed), and more subtly it has the effect of making developers more reluctant to be first adopters, since they start feeling first adopters have to pay a higher cost, and it makes authors feel like the specs aren't really worth anything because they keep changing. Plus, of course, there's the opportunity cost: making a minor improvement means we're spending lots of resources (speccing, implementating, testing, documenting, advocating) that we could instead be spending on making something else a _lot_ better. (The number one way of making a spec fail is to ignore backwards compatibility, of course. Which in a way is the same thing, just on a larger scale.) -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 7 June 2012 20:32:24 UTC