There are various different ways to test whether features work not so well.

Not in the TC case... anyway, I'm not really here to advocate the best way to do so. I'm sorry that the focus has been moved to the browser detection. I really hope some nice workaround suggestion could be found for the selection problem instead. Thanks for your comments Mike. Fred

-----Original Message----- From: Mike Hommey [mailto:mh@glandium.org] Sent: 22 September 2007 12:17 To: Frederico Caldeira Knabben Cc: 'Mark Rowe'; webkit-dev@lists.webkit.org Subject: Re: [webkit-dev] Editable selection inside empty element

On Sat, Sep 22, 2007 at 12:11:27PM +0200, Frederico Caldeira Knabben <fredck@fckeditor.net> wrote:

...

Thanks for the reply Mark. I've filled the following bug for it:

http://bugs.webkit.org/show_bug.cgi?id=15256

...

On an unrelated note, // For Safari and Opera, we must ensure the focus. if ( (/safari/i).test( navigator.userAgent ) )

It is almost never correct to test for "Safari" specifically. WebKit is what you should be aiming to detect. This will ensure that your code functions correctly in other WebKit-based browsers, such as OmniWeb, Shiira, NetNewsWire, etc.

That was just useful for the TC. But you are right; I'll take care of using the recommended way:

(/ AppleWebKit\//).test( navigator.userAgent )

Actually, this is as bad as testing for safari. You should test features, not user agents.

Mike