[standards-jig] Concern About JEP-0082

Evan Prodromou evan at prodromou.san-francisco.ca.us
Wed May 14 22:18:06 UTC 2003

So, I'm writing with a bit of concern about JEP 0082.

Two of the listed protocols that would benefit from this JEP are
XML-RPC-based (0009 and 0075). XML-RPC has a dateTime.iso8601 type for
transferring datetime values.

>From the name, you'd guess that you could put any old ISO-8601 date in
there, and it should be handled correctly. But XML-RPC is a really
imprecise protocol (having implemented it twice now, I'm not
particularly a big fan of Dave Winer's specification style), and the
only real spec on the datetime format is one "example" value:


Sadly, most XML-RPC implementations use this braindead format
(YYYYMMDDTHH:MM:SS). It's _terrible_: it leaves out timezones, it
doesn't allow millisecond granularity in times, doesn't do
nothing. But, de facto, this has become "the" datetime format for

The date profile defined in JEP 0082 is definitely superior. However,
using it for JEP 0009 and 0075 would mean breaking any compatibility
with existing XML-RPC libraries or code. I'm not sure that that's
worth it.

So, I'm suggesting that when there's an external compatibility
problem, as with XML-RPC, that the compatible date format be used
instead of JEP-0082's.

JEP-0075 is a special case, in that it has stuff that should be pretty
much drop-in XML-RPC compatible (method calls), but there's other
stuff, like attribute values or the timestamps on certain messages,
that could theoretically be switched over to JEP 0082 format without
much compatibility trouble.

However, I'd be a little queasy about doing this, as it'd mean two
different date formats in one spec. I'd think it'd make more sense to
have all dates in the same protocol have the same format, even if it
is XML-RPC's broken and stupid one.


P.S. Yes, there are problems with the dates used in JEP-0075; this will
be rationalized in the upcoming revision.

Evan Prodromou
evan at prodromou.san-francisco.ca.us

More information about the Standards mailing list