Re: [dispatch] SIP-CLF: text versus binary

["Vijay K. Gurbani" <[e-mail]>]

Adam Roach wrote:
 > Vijay's initial implementation was pretty much spot-on, with one
 > easily fixed exception.

Adam: Thanks, and once again, I take the blame for not paying
enough attention to the binary CLF write function (but that is
the whole reason why I published the source as well -- just in
case I did not account for something egregious.)

 > In either case, the speed of processing one versus the other stands at
 > 5x to 10x by any reasonable metric.

To be sure, you are saying that a perl script finding a certain
record in an ASCII CLF file is about 5x to 10x times slower
than a C program or perl script finding the same record in a
binary CLF file.  If so, yes, that is so.  The writes are
about even, with binary writes taking up a bit less time.

There are other technical discussions that I will like to
continue with you on the merits and demerits of ASCII vs.
binary, but will wait for a short while until we decide to
see how the SIP CLF work progresses in dispatch (hopefully we
can conduct those on a separate technical discussion list.)

Ciao,

- vijay
--
Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent
1960 Lucent Lane, Rm. 9C-533, Naperville, Illinois 60566 (USA)
Email: vkg@{alcatel-lucent.com,bell-labs.com,acm.org}
Web:   http://ect.bell-labs.com/who/vkg/