Re: A Compromise to the Versioning Debate

Henri Sivonen wrote:
> Why would designed-for-ie-version='7' be better than e.g. tested-with- 
> current-browser-versions-on='2007-04-16'?

   Because:

1) No one will remember what date the browser with the bug came out on.

2) The date is meaningless if you don't know what browser it applies to.
If a page depends on a bug in IE7, the date won't tell you that.

3) Because the date is meaningless outside the context of a browser,
Internet Explorer will be assumed and the attribute ends up being a
really poor way of saying "designed-for-ie-version='7.0'".

   As a result of all of the above, we need an attribute that describes
the browser and the version rather than a point in time. Perhaps
something like this:

| <html bugmode="IE7"> [...] </html>

   When IE8 comes out and some pages break, Microsoft tells everyone
they can fix it by just adding the |bugmode| attribute. In fact, they
can do so preemptively. It's not ideal, but doing the opposite and
assuming bugs unless a switch is used has two serious problems:

1) There's no explicit information indicating that the page depends on
any bugs.

2) Standards-compliant pages that are conforming before a switch is
introduced will suddenly break because they don't account for the bugs
introduced after the specification they comply with.

Received on Tuesday, 17 April 2007 02:39:44 UTC