[whatwg] IE/Win treats backslashes in path as forward slashes

[Ian Hickson]

On Wed, 11 Apr 2007, Geoffrey Sneddon wrote:
>
> Looking through the spec again, there is nothing about backslashes in 
> URI's path being treated as a forward slash, behaviour needed for 
> compatibility for quite a few websites.

On Wed, 11 Apr 2007, Gervase Markham wrote:
> 
> I would be rather surprised if that were true, given that Firefox 
> doesn't do it and I've never come across a website which broke for that 
> reason. But maybe I live a sheltered life.

On Wed, 11 Apr 2007, Bill Mason wrote:
> 
> It's not that unheard of, though I wouldn't say it's rampant.  Just a 
> quick search on bugzilla.mozilla.org [1] produces some samples.
> 
> Admittedly most of these bugs are old. However the newest one is from 
> January 2007, so the problem still crops up.
> 
> [1] http://tinyurl.com/2evdox

There are quite a few of these, even more if you start looking further in. 
In fact this bug:

   https://bugzilla.mozilla.org/show_bug.cgi?id=64488

...has 49 duplicates just to itself.


I've added a postprocessing step to the URL resolving algorithm that 
concerts all \ characters into / characters.


On Wed, 11 Apr 2007, Maciej Stachowiak wrote:
> 
> Besides the backslash thing, there are a number of URI processing rules 
> that browsers must follow for web compatibility which are either not 
> required by or directly contradictory to the URI RFCs. Documenting these 
> and fixing the relevant RFCs would be a valuable goal, but possibly 
> beyond the scope of WHATWG.

Could you elaborate on this?


On Thu, 12 Apr 2007, Julian Reschke wrote:
> 
> It seems to me that at least this thread does not point out bugs in 
> RFC3986 or RFC3987, but problems in user agents that do not follow these 
> specs. Or stated otherwise: in reality, URIs in HTML documents are not 
> RFC-compliant URIs or IRIs, but something else. It's up to the working 
> group to either deprecate these kinds of references, or to specify how 
> they should be handled.

Done.


On Wed, 11 Apr 2007, Michael A. Puls II wrote:
> 
> However, we can't specify this for all URIs (just saying). Flipping raw 
> backslashes (even though they should really be encoded) in <a 
> href="mailto:uridata"> for example, should not be done.
> 
> If we do specify this, we have to be more specific than "path" because 
> 'path' does not necessarily mean URI.

Ok, done.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'