> Ahh of course, the problem is not code migration, but data 
> migration, hmmm. 

Hmm indeed!

 
> I tried to find some code of how this had been done on earlier 
> occasions of 
> upgrading the Mailet API, but without luck. Is this really the 
> first time the 
> serialization format of Mail has changed?

Not exactly, but in principle yes. 
This time we have to acknowledge that there is a large userbase who have expressed 
concern when upgrading James already about minor incompatibility between repository 
versions, without the API changing.

> Okay, what about this approach to readObject/writeObject, that 
> would solve the 
> problem in a way eliminating the need for a data migration tool:


If it works I see no reason why not to do it this way, So.. if you can produce a 
method by which mail in existing repositories can be converted on-the-fly Good! go 
right ahead.

The way things work are that your changes will be reviewed by some of the comitters to 
make sure they are _basically_ OK, then they'll be comitted, and can be modified as 
time goes on. In other words its a collaborative process.

Therefore basic functionality is OK for the HEAD if it is sensibly designed and 
appears to work, wrinkles can be ironed out in the longer term.

d.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to