Re: [Straw-devel] Re: link/guid behaviour, feedparser changes and tracking upstream

[Juri Pakaste]

(I'm assuming you wrote this after we talked about it on IRC, at least
it seems so to me from the timestamp)

On Mon, 2004-06-28 at 22:31 +1000, Jan Alonzo wrote:
> "Some RSS feeds use guid when they mean link. guid can also be used as an
> opaque identifier that has nothing to do with links. If an RSS feed uses guid
> as the entry link and no link is present, Universal Feed Parser detects this
> and makes the guid available in d.entries[i].link."
> 
> My guess is if isPermalink="true" AND link is present, it won't overwrite
> link. It will only put the guid there IF (1) isPermalink="true", and (2) link 
> is
> NOT PRESENT.

That's right, it won't overwrite link in that case. But then we have *no
way* of knowing if e.id is a link or not, it'll just be an opaque
string. We lack the knowledge that the presence and value of isPermalink
gives.

Possible combinations and what feedparser as distributed gives us
(remembering that isPermalink="true" is the same thing as no isPermalink
attribute):

1) No link, no guid: no link, no id
2) No link, guid, isPermalink="false": no link, id is guid
3) No link, guid, isPermalink="true": link is guid, id is guid
4) Link, no guid: link, no id
5) Link, guid, isPermalink="false": link is link, id is guid
6) Link, guid, isPermalink="true": link is link, id is guid

You can't tell the difference from outside between cases 3, 5 and 6. In
case 6, we should show both link and guid links, assuming they are
different. In case 3, they'll be the same, so no problem there. In case
5, we should show only link, but we can't know that. id can be
syntactically a completely valid URI and still result in 404 when the
user clicks on it. And I believe that's a valid situation, even if I
don't know of a feed that works like that.

-- 
[ Juri Pakaste | address@hidden | http://www.iki.fi/juri/ ]

signature.asc
Description: This is a digitally signed message part