Re: Browsers breaking content negotiation

Ruben,

why are you inferring the format from the URI? The URIs are opaque;
only RDF specifies the format here.

Let me improve the example -- no filename extension, just some document URI:

  <http://some.com/img/19f87c54-bb97-4aa7-8163-166e3858f45e>
dct:format "image/jpeg" .

It identifies a resource which may have multiple representations:
JPEG, RDF, HTML etc. The representations could be standalone
resources, but they don't need to.

On Thu, Feb 9, 2017 at 7:59 PM, Ruben Verborgh <Ruben.Verborgh@ugent.be> wrote:
> Hi Martynas,
>
>> <http://some.com/img/abc.jpg> dct:format "image/jpeg" .
>
> So this URI points to a specific representation of the ABC image,
> namely a JPEG version of it.
>
>> This allows my Linked Data
>> platform to respond in at least 3 different ways to a request to such
>> URI, depending on the Accept request header:
>
> No, I don't think so.
>
>> 1. the JPEG image for image/*
>
> Correct.
>
>> 2. the RDF metadata for any RDF syntax
>
> No, that's another document;
> the image is different from the document describing it.
>
>> 3. HTML rendering of the metadata for text/html
>
> Idem.
>
>> Yet recently I noticed that some browsers started sending Accept: */*
>> instead of image-specific media types:
>
> And that's not a problem.
> With the triple above, you have defined the thing
> as a JPEG representation of something.
> So it can only be JPEG.
>
>> This provides no information for the content negotiation algorithm and
>> leading to a random response format.
>
> It will be JPEG.
>
>> Anyone else thinks such behavior breaks WWW architecture?
>
> Not me.
>
>> Browser
>> vendors apparently have decided that conneg is bad:
>> https://wiki.whatwg.org/wiki/Why_not_conneg
>
> That's a rather unfortunate one-sided argument.
> The reasoning is not sound in various places.
>
> Best,
>
> Ruben

Received on Thursday, 9 February 2017 19:06:47 UTC