Tim Bray wrote:
On Apr 25, 2005, at 11:01 AM, Graham wrote:
Can you post some links to examples of feeds you think are difficult to express in the current syntax? That would be considerably more constructive than whatever the hell that was.
What Rob wants is what he said in http://www.imc.org/atom-syntax/mail-archive/msg14157.html
I decided it would help if there was an actual Pace: http://www.intertwingly.net/wiki/pie/PaceOptionalSummary
Thanks.
Here's a meta comment, based on several years of real world implementation experience with the IETF WebDAV specs:
It doesn't make sense to REQUIRE protocol elements for which there are valid reasonable use cases where implementations can't supply them and which aren't absolutely needed for interoperability. Also, if a protocol element is indeed not absolutely required, make sure that it's clear that leaving it out is *better* than supplying bad data.
I'll supply two examples:
#1: DAV:displayname property
Many readers of RFC2518 came to the conclusion that a WebDAV resource MUST have a DAV:displayname property. Many implementations that aren't able to provide any additional metadata besides what they get from the file system thus use the filename (minus path) instead. This is not only a complete waste of bandwidth, but also causes lots of interop problems when resources like these are copied to other servers where DAV:displayname indeed works as designed (for instance, renaming the resource will not cause DAV:displayname to change anymore like it did on the origin server).
#2: DAV:getcontentlanguage property (same for HTTP "Content-Language" header)
I've seen servers providing the server installation's locale name as a default. I've seen clients (MS Webfolder) that send a Content-Language header hardwired to "en-us" for any document they HTTP PUT. In both cases, recipients receive bad metadata, instead of no data at all which would be preferrable.
In both cases, there'd be fewer interop problems (and incorrect metadata) if the specs would allow not to send these protocol elements (and be clear about that you shouldn't make them up if you don't have them).
Please consider this when discussing this Pace and the link/alternate issue as well.
Best regards, Julian
