RE: URIs, deep linking, framing, adapting and related concerns

> I very much agree with the TAG concentrating on documenting conventional 
> expectations, based on the architecture and design of Web technology. 
> However, I'm somewhat afraid that "Conventionally" could easily be read 
> as "assumption unless overwritten". I think you did address that to 
> quite some extent in the immediately preceding paragraph, but I think we 
> should do whatever we can to make the wording clearer. As an example, 
> rewording that paragraph as follows might help:

> Someone discovering such a link cannot be expected to check any "terms 
> and conditions" on the use of such a link.

"cannot reasonably be expected"; there are those who might
have an unreasonable expectation.



>> Conventionally, however, someone making a web site that offers images
>> for use in its own compound documents (for use in transclusion)  does not
>> expect others to point to or cite those resources directly; a resource mounted
>> for viewing stand-alone is different from one mounted for viewing only
>> within the site's own context.  Perhaps the content needs context,
>> does not stand alone, or perhaps it is merely because the cost of mounting
>> the image/script/resource/style-sheet is funded by whatever is behind
>> the context, and taking the resource out of context does not provide
>> a way for cost recovery.

> I'd avoid any references to costs and cost recovery, because some people 
> or organizations arguing against 'deep linking' (for a@href) also argue 
> with costs.

I think costs and cost recovery are central to the discussion, and avoiding
mentioning them would make the document pointless. It costs the newspapers
money to pay reporters to write articles, and they want to recover the cost
by making people go through the advertising click-through gauntlet, for
example.  But there are reasonable and unreasonable ways of accomplishing
the goal, and disallowing deep linking to internal stand-alone pages
that are not protected by access control doesn't seem to be reasonable.

> One point I'd make is that in contrast to technologies such as office 
> productivity tools (Microsoft Office, Open Office,...) and print-ready 
> formats (PS, PDF), the core document format of the Web (i.e. HTML) 
> includes images only by reference, without any way of doing so by value. 

The premise is wrong; some authoring formats definitely do allow and support
transclusion; PostScript and PDF also allow it although it isn't common
with PDF today. Others have pointed to "data:"; also, with MHTML (and likely 
with the webapps zip-based container format), it is also possible to 
include images in a single retrieval. MHTML is not common with HTTP
but it is with emailed HTML.

Consider the use case of emailed HTML with references to embedded
images; is it reasonable for someone to reuse those URLs in a popular
web page?

> Therefore, images available on the Web have to be assumed to be intended 
> for transclusion, not to stand alone. 

Well, I think "have to" is too strong. It's unreasonable to assume
images are intended for stand-alone use or third-party transclusion,
though.

> As for scripts and styles sheets, 
> first of all these don't serve much of a purpose purely on their own, 
> and second, there are strong arguments such as consistency and caching, 
> for using them by reference rather than by value (in this case, both is 
> possible technology-wise).

I'm not arguing to change usage, just describe it.   If I put up a web site
using my cool new style sheet and you like my style sheet and YOU want to
 use the same one, I don't really expect you to use my style sheet without 
any of my content... 

Say, for example, you wanted to put up a document that looked just like
it was officially from W3C (or CNN or whatever). Is it reasonable for you
to just use the W3C style sheet (and images, etc?). 

Larry
--
http://larry.masinter.net

Received on Thursday, 30 December 2010 20:30:28 UTC