RE: well-formedness error

["jt" <jt@xxxxxx>]

| -----Original Message-----
| [mailto:owner-atom-syntax@xxxxxxxxxxxx]On Behalf Of Walter Underwood

| --On Wednesday, June 16, 2004 12:59 AM +0200 Danny Ayers
| <danny666@xxxxxxxxxxx> wrote:
| >
| >>    [RFC 3470 on receiving bad XML ...]  Reasonable behaviors in
| >>    such a scenario could include attempting retransmission or aborting
| >>    an in-progress session.
|
| But an HTTP GET of a feed does not have retransmission or a session.
| There is no way to report the failure back to the software that sent
| the badly-formed XML.

I'm too lame to comment.. however...

| That is why I said there were two situations. One where it is possible
| to report the error within the protocol and one where it is not.
|
| I would like to go with RFC 3470 and forbid interpreting badly-formed
| XML, but the horse is already out of the barn on that one. Clients
| are doing that already. Realistically, we should specify that when
| they do it, they must log enough info to fix it and tag the data
| as less reliable than your average blog entry (that's a concept!).

If you log the entry as validated, *and by what brand/version of validator*,
you'd surely have enough info to both fix most errors and report errors,
right??

An entry that logs NOT validated can then be ignored, entries that CLAIM to
be validated but found NOT to have been can then be put in a "Hall of Shame"
website, along with the non-validated feeds.

Makes explicit, what is implicit.

Discrepancies between validators can then surface MUCH easier, as a no-small
"side-benefit".  Actually, imv, that's one-a the MAJOR benefits of this
approach (but I can be snowed by technical jargon if ya wanna...;-)!!

Whaz the downside here??


Mebbe a better way of say the above (although, half-joking, I dunno why
ANYbody would even PUT a "Valid 'FOO'" badge on ANYthing...;-):

| -----Original Message-----
| [mailto:owner-atom-syntax@xxxxxxxxxxxx]On Behalf Of Bjoern Hoehrmann

| at least the amount of web sites that have a "Valid Foo" badge
| and a link to the W3C MarkUp Validator that are actually invalid seems
| to suggest that. I think the easier to use tools that help to conform
| the more will. And I think a tool that regularly checks documents would
| be a great help and would thus mitigate the risk that implementers have
| to deal with ill-formed content.