[whatwg] Fakepath revisited

[Eduard Pascual]

On Thu, Sep 3, 2009 at 9:29 AM, Smylers<Smylers at stripey.com> wrote:
> If one major browser implements non-standard behaviour for compatibility
> with existing content, it would have an advantage with users over other
> browsers -- those other browsers would likely want to implement it, to
> avoid losing market share.  But browsers unilaterally implementing
> 'extra compatibility' means other browsers wanting to be similarly
> compatibile have to reverse engineer the first browser -- a
> time-consuming and brittle process, which in practice often leads to
> some edge cases where the behaviour is not the same.

Currently, all major browsers implement "non-standard behaviour for
compatibility with existing content", quoting your own words, or
"quirks modes", to use more common terms. Does the HTML5 spec define
what should trigger each "quirks mode" and how browsers should behave
when in those modes? If it did, then the "fakepath" could be treated
as just another quirk, and end of the problem. But as far as I know
the spec doesn't dig too deep (correct me if I'm wrong), so there are
gonna be thousands or (quite likely) millions of sites that will break
unless browsers special-case them as they currently do with their
"quirks modes". The barrier for entry for new browser vendors is
already huge on this area, and stupidifying <input type=file> with
fakepaths will *not* solve this.