To: ietf-822(_at_)dimacs(_dot_)rutgers(_dot_)edu
Subject: Re: Massive Content-Type definition ideas & Gopher
Date: Tue, 8 Jun 1993 07:50:05 -0500
At 1:09 AM 6/8/93 -0400, Keith Moore wrote:
Some people do seem to think that using twice the bandwidth to transmit a
file is a significant amount. And for disk storage, $.50/megabyte does
seem a lot cheaper than $1.00/megabyte.
Certainly, compression seems very attractive if you ignore the costs it
incurs and only look at the costs it saves.
There is cpu and wall time spent [de]compressing, there is decreased
interoperability, there is increased complexity in the protocol; all these
things have costs, too.
Agreed. You have to consider all of these things.
The current version is 1.8 megabytes.
I'm sure you could squeeze that down by more efficient coding, removal of
extraneous features, etc. Don't you want to save bandwidth and storage?
Or are the savings you would gain outweighed by the effort it takes to make
them?
It's much easier to simply compress the source code. And I doubt I could
save 50% by using all of the techniques you mention. In the case of this
code, the user has to extract, compile, and install it, not to mention
getting it to work with the particular perversion of X windows that his
vendor supplies. The overhead of decompressing it is marginal by comparsion.
(So yes, a cost/benefit analysis has been performed, and compression won.)
Perhaps compression really is a net gain, but it's not quite so simple as
"some people think $.50 is cheaper than $1".
Not quite that simple. But it seems like compression is mostly used for
things like source distributions, where the overhead in decompressing the
file is considered to be worth the savings. I don't expect MIME compression
to be used for normal text/plain email messages.
Keith