A few points:
1. "x-" is commonly used in cases when an application for a mime type is
pending, and when there is a reasonable expectation that it will be
approved. The mime type is prefixed with "x-" until the requested mime
type becomes official, after which the "x-" is dropped.
2. We will be registering MODS and MARCXML:
3. The reason one uses (or doesn't use) +xml is made very clear in one of
the relevant RFCs (I don't have the number at the moment): the application
consuming the content is supposed to recognize the mime type and process it
accordingly, however, in the event that it does not recognize the mime type,
the "+xml" signals at least that the content is xml, and so there is a
possibility that it might do something useful with it, even though it cannot
proccess it according to mime type - it may be able to parse the XML and
present something readable to the user. Even better, consider the case
where it is a protocol response, for example SRU, where we are registering
application/sru+xml, there might be an accompanying stylesheet url, and the
client can then format a complete sru response without knowing that it did
The reason is NOT, as some have suggested, to distinguish "mods+xml" from
"mods+xyz" where "xyz" is some alternative syntax. However, because of the
confusion, we would register marcxml as marcxml+xml (even though it sounds
funny) rather than marc+xml, because of all the confusion that the latter
name would cause.
----- Original Message -----
From: "Jonathan Rochkind" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Thursday, February 12, 2009 5:21 PM
Subject: Re: [CODE4LIB] MIME Type for MARC, Mods, etc.?
> Actually, re-reading some of the RFCs, I would clarify one thing.
> It seems like using unregistered "x-" MIME type is discouraged, and
> instead you are encouraged to use what is (claimed to be) a very quick and
> easy and painless process of registering "vnd." types. So I'd encourage
> LC to investigate doing that for MARC, while waiting for someone to have
> time to do an actual (more time consuming) application/marc+xml
> registration. That would give us the beneift of an actual registration
> (albeit under vnc.) instead of an unregistered x-.
> As far as text/xml, the general consensus on the internet seems to be that
> it was a mistake, but it's there and no one cares enough to try to somehow
> remove it, so it _is_ legal, but nobody really encourages using it. One
> problem with text/html is that it's default char encoding is ascii, while
> the default char encoding for XML is of course UTF-8. This can very easily
> lead to confusion and encoding errors unless software is more careful than
> we know most software has a tendency to be. :) Still, it's legal, but I
> don't see any reason to encourage it's use for MARC.
> application/xml, sure, but it would be _really_ useful, for the reasons
> discussed in last week's thread, to have a specific type for marc xml (and
> mods). If the folks at LC don't understand why, thinking that
> application/xml is sufficient, i could try to write up a persuasive essay
> again, or copy and paste from last week's thread. Or is there someone else
> other than LC who could conceivably fill out an application for
> application/marc+xml and application/mods_xml?
> Seriously, application/xml is not sufficient, although it is legal.
> Alexander Johannesen wrote:
>> On Thu, Feb 12, 2009 at 22:32, Jonathan Rochkind <[log in to unmask]>
>>> Didn't we finish having this conversation last week? We talked about all
>>> this stuff being brought up now last week.
>> We did indeed, and your summary is better than what my retort could
>> have been; spot on.
>> I guess it's hard to understand why text/xml is such a waste of MIME
>> and time as long as we still got text/html as the original understood
>> MIME for HTML pages, but luckily the internet has moved on and
>> evolved. :)
>> One question we haven't asked is if we really need a MIME type for
>> MARCXML. :)
>> Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic
>> http://shelter.nu/blog/ --------