Path: utzoo!utgpu!cs.utexas.edu!yale!cmcl2!kramden.acf.nyu.edu!brnstnd
From: brnstnd@kramden.acf.nyu.edu (Dan Bernstein)
Newsgroups: alt.sources.d
Subject: Re: Read this if you're having trouble unpacking Tcl
Message-ID: <20210:Jan601:32:3091@kramden.acf.nyu.edu>
Date: 6 Jan 91 01:32:30 GMT
References: <1990Dec27.071632.7272@zorch.SF-Bay.ORG> <97261639@bfmny0.BFM.COM> <2852@cirrusl.UUCP>
Organization: IR
Lines: 26

In article <2852@cirrusl.UUCP> dhesi%cirrusl@oliveb.ATC.olivetti.com (Rahul Dhesi) writes:
> Agreed with everything that Tom Neff says -- I rather prefer that
> human-readable source remain human-readable when posted in Usenet
> articles.

Yeah. If tabs and control characters and so on are a real problem, we
can easily define a text encoding that's both readable and safe. What
characters do we need? !@#$%^&*()-=_+[]{};':",./<>?\|`~, space, tab,
letters, numbers, and perhaps some other ASCII control characters. At
least letters, numbers, space, and !$%&*()-=+;:",./?\ will make it
through any translation. So we just define codes for the other
characters and be done with it. A machine without ~ can give an error if
it sees the code for ~. Binary files which inherently depend on the 256
bytes can be uuencoded. Done.

> However, some sort of checksum should be included for verification.

True. A cryptographic hash function, like Merkle's Snefru with 3 or 4
passes, would suffice.

> Unfortunately most people don't include such a check code when
> they post shar'd source,

But won't all this go into the shar program?

---Dan