Re: Tool request

Sean B. Palmer wrote:
> Sam Ruby wrote:
> 
>>What would be considerably harder is convincing people to install
>>the feedvalidator on their own machines.
> 
> It might help if there were a single-file standalone version,
> especially one for each of the most popular flavours of feed that need
> validating. I notice, for example, that you include rdflib in the
> current distribution [1], which is presumably only necessary for RSS
> 1.0 and 1.1. If that's not possible, as stripped a version as possible
> would be best.
> 
> It'd also be great if there were a .tar.gz of the most recent version
> of the Feed Validator, and there possibly already is one but I can't
> find it. If so, a more obvious link from either the front page or
> somewhere close would be really appreciated.

http://www.feedvalidator.org/ => about => Can I run it locally?

> What I'm envisioning is being able to go to feedvalidator.org, follow
> one or two links to a tarball, inflate it, move it to a web directory,
> and have it Just Work.

What's not mentiond on the about page (and should be) is that ther are
.tar.gz files for you to play with at <http://feedvalidator.org/download/>.

Click on today's date.

Peek at the README, and you may need to edit config.py.  Let me know how
it goes.  There isn't much in config.py, and most of what is there
doesn't need to be modified.

> As for Danny's idea of having Feed and RDF Validator logo services,
> one obvious point that hasn't been mentioned yet is the fact that
> referer-based validation won't work there (except on XSLT'd RDF; FOAF
> profiles and the like), so the URI would have to be passed at a GET
> query instead. This is actually more interesting than my current
> referer based approach since a) you can see the validation status of
> any page on any page, so you could have a huge list of logos
> somewhere; and b) it'll work even in browsers that fake or chop or
> otherwise mess up the referer header.

Autodiscovery can make validating the referer possible (and
autodiscovery logic is checked into the feedvalidator), but yes, the
interface could also accept a URI to validate.

> I'm also experimenting with more flexible approaches to validation,
> using custom subsets and RELAX NG and the like [2], to take away the
> burden of centralised services. Terje also keeps teasing me with
> OpenSP, saying that it would enable a standalone W3C Validator, which
> would be great--I'd really like a small command line wrapper for it,
> along the lines of what Björn's already started to experiment with.
> Since OpenSP 1.5.2 was released last month [3], I wonder if anybody's
> got cracking on this?

FYI: the existing feedvalidator already has a simple command line wrapper:

  $ python src/demo.py http://miscoranda.com/feed
  Validating http://miscoranda.com/feed
  No errors or warnings

- Sam Ruby

Received on Sunday, 29 January 2006 13:38:44 UTC