4 ms·

what he says is still valid,the spec hasnt changed,is broken even or recent browsers such as Chrome ,is complicated and confusing.It's another exemple of half b

by aikah 12y ago

what he says is still valid,the spec hasnt changed,is broken even or recent browsers such as Chrome ,is complicated and confusing.It's another exemple of half baked spec written by people that are never going to use it,just like app cache. Frankly HTML5 is a disaster(and no , things like WebWorkers or WebSockets are not part of the HTML5 spec before anyone jumps on me). It's just hard to build something robust on weak foundations,but that's lot of client side developpers, making these things work even though specs are just not helping.

shacharz 12y ago
You're right, but that's the reason I think all browsers have developer channels. So that developers who actually user those API's will test them, report bugs or even stir the specs towards something useful. I know that's the case at least with WebRTC.
thristian 12y ago
I don't think it's fair to blame the HTML5 drag and drop API entirely on WHATWG; my understanding is that it was originally specced years ago as the sanest useful subset of IE6's drag and drop API. It may not be the prettiest thing ever, but it's prettier than how it started out.
- panzi 12y ago
  Subset? Isn't it more like a super set? IE6 (and above) has no files array that can be read with a FileReader and it only supports the types "Text" and "URL", no support for proper mime types.
  - thristian 12y ago
    I think this is what I was remembering: http://ln.hixie.ch/?start=1115899732&count=1 http://ln.hixie.ch/?start=1115899732&count=1 ...which is to say, IE6 supported more event types and more configurable behaviour than HTML5 supports, but some of those extra features were just too weird, or outright crashed IE, so they didn't make it into the spec.
mschuster91 12y ago
> It's another exemple of half baked spec written by people that are never going to use it,just like app cache. Appcache has ever worked flawlessly for me. IndexedDB, though, is a the feces of a dog who has eaten too much cow shit. Why not WebSQL? Shit, even if the only implementation used is SQLite, it's better than IndexedDB which is asynchronous, can't even do a motherfucking sort, a simple enough range query or unions. Fuck all that NoSQL crap, thank you very much.
- panzi 12y ago
  The problem was that SQL is a giant standard that nobody implements really correctly and it would blow up the HTML5 standard enormously. Also I don't think there is anything wrong with an async-only API that does potentially do disk access.
  - mschuster91 12y ago
    I'm fine with the async part, if there was a proper support for stuff like joins and sorts. Without this, you have to individually gather the single rows and do it in userspace code. This is crazy enough, add async-ness to the whole mess and even with promises you end up with a shitload of spaghetti code. Right now IndexedDB is a kids playground and nothing usable at all for any serious application.
ulisesrmzroche 12y ago
It's changed a lot, try finding excerpts of the article (the quotes) on the new draft. http://www.w3.org/html/wg/drafts/html/CR/editing.html#dnd http://www.w3.org/html/wg/drafts/html/CR/editing.html#dnd

by aikah 12y ago

shacharz 12y ago

You're right, but that's the reason I think all browsers have developer channels. So that developers who actually user those API's will test them, report bugs or even stir the specs towards something useful. I know that's the case at least with WebRTC.

thristian 12y ago

I don't think it's fair to blame the HTML5 drag and drop API entirely on WHATWG; my understanding is that it was originally specced years ago as the sanest useful subset of IE6's drag and drop API. It may not be the prettiest thing ever, but it's prettier than how it started out.

panzi 12y ago
Subset? Isn't it more like a super set? IE6 (and above) has no files array that can be read with a FileReader and it only supports the types "Text" and "URL", no support for proper mime types.
- thristian 12y ago
  I think this is what I was remembering: http://ln.hixie.ch/?start=1115899732&count=1 http://ln.hixie.ch/?start=1115899732&count=1 ...which is to say, IE6 supported more event types and more configurable behaviour than HTML5 supports, but some of those extra features were just too weird, or outright crashed IE, so they didn't make it into the spec.

mschuster91 12y ago

> It's another exemple of half baked spec written by people that are never going to use it,just like app cache. Appcache has ever worked flawlessly for me. IndexedDB, though, is a the feces of a dog who has eaten too much cow shit. Why not WebSQL? Shit, even if the only implementation used is SQLite, it's better than IndexedDB which is asynchronous, can't even do a motherfucking sort, a simple enough range query or unions. Fuck all that NoSQL crap, thank you very much.

panzi 12y ago
The problem was that SQL is a giant standard that nobody implements really correctly and it would blow up the HTML5 standard enormously. Also I don't think there is anything wrong with an async-only API that does potentially do disk access.
- mschuster91 12y ago
  I'm fine with the async part, if there was a proper support for stuff like joins and sorts. Without this, you have to individually gather the single rows and do it in userspace code. This is crazy enough, add async-ness to the whole mess and even with promises you end up with a shitload of spaghetti code. Right now IndexedDB is a kids playground and nothing usable at all for any serious application.

ulisesrmzroche 12y ago

It's changed a lot, try finding excerpts of the article (the quotes) on the new draft. http://www.w3.org/html/wg/drafts/html/CR/editing.html#dnd http://www.w3.org/html/wg/drafts/html/CR/editing.html#dnd