SOMETIMES POST COMPRESSED UUENCODED TAR FILES!!!
Jef Poskanzer
jef at ace.ee.lbl.gov
Sun Mar 26 08:53:15 AEST 1989
In the referenced message, perry at bellcore.com (Perry E Metzger) wrote:
}My apologies for posting this to alt.sources instead of
}comp.sources.d, but hell, no one reads comp.sources.d, and this is
}important.
"Good excuse."
}DONT POST UUENCODED COMPRESSED STUFF.
}
}Compress gets run on most stuff crossing the net,
This is becoming less and less true. NNTP links do not use compress,
but they do represent the fastest growing segment of Usenet traffic.
But even on systems where transmission is compressed, storage is not.
In most cases it doesn't make any significant difference either way,
so it's better not to compress&uuencode for ease of perusual. However,
there have been a few postings within the past year that definitely
should have been compressed&uuencoded. The Mahjonng tiles. The
Laserjet fonts. The SAO star data. These datasets all had large
compression ratios, and if stored uncompressed they would have used
up significant space in the net's collective spool directories. And
furthermore, since they were data, no one was likely to want to
casually peruse them.
My guidelines for whether to compress&uuencode a posting:
- Is it huge ( > 500K)?
- Does it compress well ( > 5:1)?
- Is it of limited *casual* interest?
By the way, if someone wants to bring up the issue of compressing already
compressed stuff, please do some actual experiments first. As in, size of
X, size of compressed X, size of double-compressed X, and size of compressed
uuencoded and compressed X. You may be surprised at the results.
---
Jef
Jef Poskanzer jef at helios.ee.lbl.gov ...well!pokey
"It's the real thing."
More information about the Alt.sources
mailing list