11

It seems to be common (at least among my section of the world's *NIX users) to refer to "tape archives" made with GNU tar or similar as "tarballs". I always thought this was a slightly odd choice, but the imagery does kind of make sense.

The other day I was reading about S-100 systems and how they worked, and came across a note that one of the the most popular tape-drive controller cards for the home enthusiast was the "tarbell"...

A one-vowel difference seems like the kind of thing that couldn't just be a coincidence...

Is there any evidence for or against the name of the modern tarball archive being descended from the name of the once-popular tarbell tape drive controller? Or did the influence perhaps run in the opposite direction?

  • 3
    FWIW, The Jargon File claims it's 'prob[ably] based on the “tar baby” in the Uncle Remus folk tales'. I imagine whoever wrote that was unfamiliar with either tar balls or tar babies. – Kelvin Sherlock Sep 19 '18 at 23:54
  • 3
    Bear in mind that a "tarball" is also a physical thing: en.wikipedia.org/wiki/Tarball_(oil) . I believe the term pre-dates modern computers. It's possible that somebody heard the petroleum-related term and applied it to .tar archive files. – fadden Sep 20 '18 at 15:14
  • I concur - 'tarball' is a common word for me (I grew up in a beach town after an oil tanker had gone aground off the southwestern coast of England). Given that 'tar' was the name of the tape archive program, it seems but a small step to refer to its output as a tarball. – dave Sep 20 '18 at 22:33
10

Tar stood for 'tape archiver' so there is no real doubt where the 'tar' name itself came from. The V7 papers and manual pages are fairly explicit on it's origin replacing v6 tp.

The other problem with the idea is that 'tarball' appears to be relatively modern as a term. I can't find a book reference before 1997 that uses the word 'tarball' - possibly not surprising since the use of 'tarball' in environmental science is also relatively modern. It was used on Usenet a little before (Clay Luther used it on comp.sources.unix in 1994) but on a trawl of Usenet I can't find anything pre 1990. Mid 1990s yes and by 1997 it's mainstream because FreeBSD used the term a lot.

The Jargon File also agrees mainstream use was 1990's and suggests it came from tar-baby but as is typically the case with that document offers no evidence for the claim of any kind.

So although we think of 'tarball' as an old old word - it's not.

5

Any naming relation is very unlikely.

The inventor and manufacturer of the Tarbell cassette interface simply had that name - I pretty much doubt he had his name changed for such a small pun.

The tarball, however, derives from Tape ARchive, a very obvious acronym. No dependency from this direction as well.

Tarbell was a name used in the computer hobbyists scene, while Unix tar thrived in the professional world. Even if both denominations first occured in roughly the same time (mid-end seventies), this looks like a real coincidence to me. Also, the technology used in those two worlds was rather different: The Tarbell interface connected to simple audio cassette recorders and wasn't even able to store a (then, professional Unix) tarball (pun intended). The computers that had a Tarbell interface fitted typically also had this as the only storage media and normally even lacked a floppy disk drive.

  • I'll admit that I've never seen a Unix system equipped to store data on audio cassettes, and I don't recall seeing tar used on a non-Unix machine until long after the end of the cassette era; but a tarball is just data—a sequence of 8-bit bytes—and audio cassettes were used to store 8-bit data, so there's nothing, in principle, that would have stopped somebody from storing a tarball on an audio cassette if they'd wanted to do. – user10478 Sep 20 '18 at 14:56
  • @besmirched Not impossible, but unlikely - tar has always been Unix, and computers running Unix were highly unlikely in the environment that had Tarbell interfaces fitted. – tofro Sep 20 '18 at 21:19
  • 2
    I think we both missed an important point. It's not really about "Unix people did it this way and hobbyists did it that way." The bigger issue was that the entire point of tar was to make archive copies of collections of files from a computer hard drive. While, the whole point of Tarbell was, you couldn't afford a hard drive. – user10478 Sep 20 '18 at 21:50
  • I was aware there was a significant budget difference between the two. It just doesn't seem that far-fetched that someone coming from a hobbyist background and moving to professional might habitually use "tarbell" as a general term for tape archives and that would then get misread/heard as "tarball" by people who didn't start as hobbyists and had never used the cassette interface card... It's only about 15 years between when the card first came out and when tarball started showing up on Usenet... – Perkins Sep 24 '18 at 22:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.