W3C home > Mailing lists > Public > public-webapps@w3.org > July to September 2009

Re: [File API] events vs callbacks

From: Olli Pettay <Olli.Pettay@helsinki.fi>
Date: Wed, 12 Aug 2009 02:39:27 +0300
Message-ID: <4A82012F.5000708@helsinki.fi>
To: Garrett Smith <dhtmlkitchen@gmail.com>
CC: Jonas Sicking <jonas@sicking.cc>, Anne van Kesteren <annevk@opera.com>, arun@mozilla.com, Web Applications Working Group WG <public-webapps@w3.org>
On 8/12/09 2:18 AM, Garrett Smith wrote:
> On Tue, Aug 11, 2009 at 2:25 PM, Olli Pettay<Olli.Pettay@helsinki.fi>  wrote:
>> On 8/11/09 11:57 PM, Jonas Sicking wrote:
>>>
>>> My concern isn't that there are ways of using it correctly, my concern
>>> is that it's very easy to use incorrectly with bugs as a result.
>>
>> How? Especially if we prevent more than one read at time. How is the
>> situation any worse than with XHR?
>>
>
> As stated, anyone can request a resource via XHR at anytime. Old CGI
> progs were single threaded, but modern programs are multi-threaded
> (e.g. Servlet), there is no prerequisite to ask the server if a file
> read is OK first. Caller makes an XHR and it either succeeds or
> doesn't. XHR doesn't require expectation on the caller; what state is
> the file in.
But if some other caller has used the same XHR, the request may be 
in-process and making a new request cancels the previous one.
I think FileData could have similar behavior.
Getting the actual data from FileData is the "request". It succeeds or 
not. Just like with XHR.

>
> Besides a file shouldn't be having behavior (read, write, delete).
Why not? (This is really a question :) )

> Insted it hould be just having some properties like "name" or "size".
> A good reason for having a separate object for read().
Well, sure, if there is a use case, File could be a separate object and
not extend FileData.

-Olli
Received on Tuesday, 11 August 2009 23:40:21 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:49:33 GMT