With the rapid spread of the global Internet, which by 1991 reaches more than a half-million computers¸Internetsize of all around the world [#!Lottor:CACM-34-11-21!#],¸Lotter, Mark the opportunities for free exchange of software and textual data are greatly enhanced.
While this brings exciting new capabilities to many people, not just those involved in academic research, it is hampered by several factors.
First, not all network file exchange¸network file exchange is error-free. Electronic mail¸electronic mailcorruption problems systems in particular are notorious for corrupting information, either by truncation of lines or message bodies, or by transliteration or other altering of certain characters. These problems are most severe for mail exchanges between major networks, such as between the Internet and Usenet¸Usenet or Bitnet.¸Bitnet
Second, no standards yet exist for describing the contents of files. While this is an area of research at some academic institutions, the wide variety of operating systems in use, and the growing numbers of computers (approaching 100 million on a world-wide basis in 1991), suggest that such standards may never exist, any more than products on the commercial market, from soup to saltines, have standard labels.
Third, without a record of origin of software and data, it is impossible for users to verify that they have up-to-date copies, or to contribute improvements and additions back to the original authors.
Fourth, without a standard means of encoding information in file headers, there is no hope of automating the process of collecting information from file headers to produce enhanced file archive summaries, catalogs, and the like.
During the author's 1991-92 tenure as President of the TEX Users Group,¸TeX Users Group efforts were undertaken to improve the quality and quantity of electronic distribution of TEX-related software and data. While this work had a narrow focus, it has quite general ramifications, and the GNU Emacs support code here is quite general, and capable of handling almost any type of computer-readable textual material.
It does not , however, address the issue of exchange of binary (non-textual) data; that has a number of difficulties associated with it, the two most severe being rigid formats intolerant of extension, and machine-specific encoding and byte order.
During a visit to Heidelberg University¸Heidelberg University in June 1990, the author spent a pleasant brain-storming session that lasted until 3am with a dozen colleagues (who names, alas, were unrecorded) from Heidelberg, Mainz, Darmstadt, and Goettingen.
We discussed many things that evening, but one topic in particular led to this work: an informal proposal for standard file headers that could address all of the problems noted above.
What's in a header?, Putting it all together, Background, Top