mangler@cit-vax.Caltech.Edu (System Mangler) (07/24/87)
In article <800@nu3b2.UUCP>, rwhite@nu3b2.UUCP (Robert C. White Jr.) writes: > 1) Sources and Binaries would be "attached" to messages rather than in > them. This would be done by a headder line very similar to the Refferences: > line. The presence of one or more "Source: <####@foo.UUCP>" or > "Binary: <####@foo.UUCP>" in the text message would indicate the presence > of that/those "Control: Source" or "Control: Binary" messages listed. This idea ought to be generalized to be analogous to the "enclosures" attached to ordinary paper mail. An enclosure is simply some material that the author of the message wanted to present verbatim without adding any letterhead or other junk to it. Since an enclosure would be separate from the article, it need not necessarily be subject to the transformations that articles must withstand, for instance, it shouldn't need any headers of its own. Keeping the enclosure "pure" would obviate most of the effort currently spent in packaging machine-readable files to survive the molestations of news transport. Instead of encoding a binary, just transmit it as is! Instead of shar files, have one enclosure per source file. There are more things than sources and executables that one would like to attach to messages, such as graphics and tar files. They'll compress better, and not require any unpacking to use, if they can be transferred as a file rather than like a mail message. The quotes that we all put in our messages to provide context could also be done as an enclosure, to get them out of the way (and perhaps reduce duplication by letting several articles refer to the same enclosure). Don Speck speck@vlsi.caltech.edu {ll-xn,rutgers,amdahl}!cit-vax!speck