On 28 Mar 2009, at 18:30, John Panzer wrote:

So people can see what we're talking about easily, I exported a (test) blog
and uploaded the xml here:

http://www.johnpanzer.com/blogexport.xml

Unfortunately this one didn't have comments, and only two entries, so most
of the data is taken up with settings.

What led you to actually try and use Atom to represent settings? It would seem far simpler to use some custom namespace for them, with a vocabulary specific to them — settings are more-or-less tool specific, so you gain little by having them in some format such as Atom.

If you view it in a feed reader, you
get something that looks a bit like key-value pairs (where each entry is a key with a description and the content is the value). There are various drawbacks to the format, I'm not proposing it as a defined standard, but we
needed something to allow for export and import and it's an (IMHO)
reasonable starting point.

It probably is a reasonable starting point, and having spoken to authors of a variety of blogging packages it seems going for something closer to it than what the current I-D has is more favoured (primarily because dealing with tar files in a memory efficient way can become fun, to say the least, in PHP, where there are quite often memory limits (per process) of around 8MB, and the export files could easily exceed that).

In some ways I'm tempted to try and come up with something that is compatible with existing Blogger-export-format importers, even if it isn't compatible the other way around (I think the latter would cripple the format from some realistic requirements that some packages have, whereas there is little reason not to do the former).

Finally, what are the drawbacks to the format from your point of view: you can probably think of things that I can't, in part simply because you have implementation experience with it.

Looking at it in a reader, it's pretty clear that critical information isn't being picked up by the generic reader, but it degrades fairly gracefully I
think.

I think there are one or two ways it could degrade better, but let's ignore this for now: there is no need to get into an endless paint-the- bikeshed discussion. I'll try and get another I-D together, um, sometime, though probably May/June.


--
Geoffrey Sneddon
<http://gsnedders.com/>

Reply via email to