> I made use of this existing class in order to detect the end 
> of the XMI.content and allow some listener to inject 
> XMI.extensions into the correct place in the output.

This relates to #5 in the list of issues that I sent in response to your
request for comments on your proposal to completely change the way we
persist projects.

> I'm sure the byte by byte parsing can be improved upon but if 
> you have some completely different idea of how to achieve 
> this then do let me know.

Inefficiencies aside, bytes and characters are entirely different beasts.
If you require a single byte character set with a particular encoding so
that you can check bytes (instead of characters) for a '>' then the code
probably needs to make sure that assumption is fulfilled.

As I mentioned in my previous note though, I really think: a) this is the
wrong time to be doing this and b) if it is done, this is the wrong level to
be doing this processing at.

This whole thing needs to be much better thought out.  I'm particularly
disappointed to see things like the arbitrary removal of support for
imported ActivityGraphs without any thought for the impact on AndroMDA
users.

Do the advantages of these changes outweigh the disadvantages?  I'm not
convinced.

Tom

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

Reply via email to