[Top] [All Lists]

Re: "Header Reordering", yet again

2005-05-27 15:28:37

On Fri May 27 2005 14:41, Paul Smith wrote:

So, why not have some explicit ordering of any new trace fields, rather  
than relying in their order in the header

For instance, put a sequence number in the header data

As with any proposed new functionality, good luck getting all of
the deployed infrastructure to implement it (which necessarily
involves getting users to update).

Add to that the likely case (considering the statements about
ordering made here, where one might expect people to know better)
of one or more developers assuming that he only need to examine
the first field to find the current maximum value...

There's also the philosophic/architectural objection that the
scheme requires poking about in the content, which is not supposed
to happen in SMTP (with the notable exception of gateways).

If you were to ignore that objection, you'd still have to specify
what's supposed to happen when a missing (as Ned noted, fields
sometimes get dropped) or duplicate sequence number is found.