- From: Ian Hickson <ian@hixie.ch>
- Date: Mon, 15 Apr 2013 05:03:10 +0000 (UTC)
- To: David Singer <singer@apple.com>
- cc: Glenn Maynard <glenn@zewt.org>, Silvia Pfeiffer <silviapfeiffer1@gmail.com>, David Ronca <dronca@netflix.com>, "public-texttracks@w3.org" <public-texttracks@w3.org>
On Mon, 15 Apr 2013, David Singer wrote: > > well, I think that there is a difference between user-comments (free > form) and machine-readable attribute-value pairs. If the data is proprietary, it's not "machine readable" is a useful sense. > also, if we have places for > > 1) pre-defined attribute-value pairs > 2) comments > > it's not obvious to me that (3) externally-defined, experimental, or > other attribute-value pairs are better in the second than the first; > they seem closer to the first, to me. reserving X- attribute names for > experiments and so on is a pretty time-honored tradition, as well. CSS and JavaScript don't have any mechanism like this, and it doesn't seem to have hurt them. HTML does have a mechanism like this, and it has proved to be a huge mistake. People waste tons of time pouring useless metadata into their documents, the data is almost uniformly bogus (hidden metadata always ends up that way), the systems most people assume are reading the data (e.g. Google) ignore it because it's so spammy and bogus, it's just terrible. I would strongly recommend against formal metadata mechanisms in WebVTT. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Monday, 15 April 2013 05:03:33 UTC