We got some great feedback from this entry on DateTime.  The dev lead for the BCL took time to post to my comments, but I thought I’d put them into the mainfeed as I think they are generally interesting.  

 

I am a colleague of Brad's and I'm the development owner of DateTime. I wanted to respond to the feedback on this thread, particularly the excellent feedback from Douglas Husemann (here) and Jeremy Gray (here).

 

Serialization in XML is the number 1 complaint associated with the DateTime. It is definitely something we are committed to fixing in Whidbey.

 

There may be a misunderstanding, but there is no current plan to introduce a new "Date" data type to the System namespace. It is important to keep the number of distinct data types in the System namespace small because so much code needs to special-case it. The DateTime class was designed to meet the scenarios associated with a Date, a Time and a combined Date and Time. There are cases in the rest of the framework where some formatting options effectively preclude the usage as a stand-alone Date which we will be correcting in Whidbey. The use of the "zzz" format by XML Serialization, XML Convert, and DataSet is the most extreme example of this because this format is only valid in cases where the time is significant and the time is Local. It precludes Date-only and UTC scenarios. If we could wind the clock back to before shipping V1.0 we would have left the "zzz" format off when serializing DateTime.

 

Regarding the request for a servicing release. Introducing new functionality into our V1.0 and V1.1 releases is always significantly more risky and expensive than a regular release. Feedback such as you are providing here is very useful in convincing people that the extra cost and risk is justified. In fact the more specific business impact and adoption impact about this issue from as many distinct sources as I can get is useful in trying to get all the right teams and individuals bought in that this work should be done. If you know others impacted by this, or you can provide more information about its impact on you, that would be helpful in making something happen. Please contact me (amoore-at-nospam.microsoft.com) if there is more feedback you can provide..

 

Regarding compatibility: the only circumstances when I see customers significantly more inflamed than you are right now about this is not when we have bugs in our product, but when fixing bugs causes reasonably functioning code to break. And compatibility issues often go directly to impacting your users and customers as they upgrade their operating systems or runtimes, or even when they install service packs. Unfortunately, it is easy to take a dependency on the weird way DateTime serializes in XML, so we have to be very careful about how we introduce fixes in Whidbey, and we have to be even more careful about anything we consider in servicing, which has an even higher compatibility bar.

 

Requests for a "historical" DateTime have been very infrequent. We are unlikely to provide such a thing in the Longhorn timeframe due to lack of demand. There may be a good 3rd party opportunity for a library for historical dating that interoperates well with the DateTime class.

 

I will pass on the request for more calendars to the team that owns that area.

 

I am not clear about your request for a more culturally aware DateTime. This FAQ may be of help in finding ways to do this with the current framework: