Re: Magic numbers again

Dror,

Sorry for the silence--as Peter Drayton points out [1], there is a
version number in the -02 draft. Although a magic number can be seen as
different from a version number, it at least give some understanding of
what it is.=20

>Does anyone of the DIME leaders care to explain the reason why no magic
>number is used in DIME. The question was rased in the past but=20
>no answer
>was given.
>
>Magic number adds virtually no overhead to the protocol (4 bypes per
>message) but it can be very helpful in various scenarios:
>
>- When analyzing wire protocols, a magic number makes it easy=20
>to identify
>the protocol and present it correctly
>- When a server needs to support multiple protocols, a magic number is
>indispensable. This is even more so with a new protocol such=20
>as DIME that
>may be used to replace older or proprietary protocols. Facilitating the
>recognition of DIME can help DIME to catch on faster.
>
>An alternative to a magic could be a common agreement on a special
>DIME 'magic record' but that's artificial and less usefull.

Henrik Frystyk Nielsen
mailto:[email protected]

[1]
http://discuss.develop.com/archives/wa.exe?A2=3Dind0207c&L=3Ddime&T=3D0&F=
=3D&S=3D&
P=3D366