Cheshire (cheshire) wrote in unixhistory,

You can tune a filesystem, but you can't tune a fish.

"You can tune a filesystem, but you can't tune a fish."

--tunefs(8), 4.2BSD and later

This appendum into the bugs section of the 4.2BSD and later tunefs manual page (I cannot verify it existed in 4BSD prior to 4.2, only that it did not exist in the 2BSD branch, and I have my suspicions that it is FFS-specific only since Research Sixth Edition UNIX did not include it and FFS was incorporated in 4.2BSD), has, to say the least, produced some curiosity with me.

So I researched it (and it would help if so many people didn't have it in their sigs).

Nick Hilliard on an ILUG mailing list on Sun Mar 12 2000 believes, via this message, that it was translated into Japanese at one point, and when reverse translated, it was "You can modify a disk, but you can't change a fish." This message was eventually removed from the Japanese man page.

According to Noah Friedman on alt.folklore.computers on 18 Mar 1992, via this thread, he hinted that the "tune a fish" sentence was removed in SunOS 4.1.0 and subsequently reinserted in SunOS 4.1.1. SunOS 4, completely Berkeley based, was replaced with the SVR4 based SunOS 5.* (Solaris), and the tunefs manpage no longer included this text. However, Friedman indicated that the reinsertion into 4.1.1 appeared as the following raw troff in the tunefs manpage:

        .\" Take this out and a Unix Demon will dog your steps from now until
        .\" the time_t's wrap around.
        You can tune a file system, but you can't tune a fish.

This man page text also has been preserved in HP-UX 7.0, although this individual hints that HP-UX 3.1 moved it from BUGS to WARNINGS.

This message proves the "tunefs" sentence has existed since at least 7 Dec 1983, which is after 4.2BSD according to McKusick. (That same thread, coincidentally, includes a reply concerning the last sentence of the BUGS section in the fed(1) manpage, namely, "The HP 2648 terminal on which this runs has been stolen").

Namely, I was trying to determine the exact time and origin of the statement, who incorporated it into the documentation, and what it meant other than the semi-obvious "you can tune a piano, but you can't tuna fish" parallel. I came across a link which indicated that the original word play was "you can tune a filesystem, but you can't tunefs" but this is unverified and I haven't been able to find anything confirming it. There is a document stating that most versions of the tunefs manpage state you cannot tune a fish, but apparently on OSF/1, you can tune a fish. They didn't include enough information to explain this, but I figure they removed the sentence.

So with all this, I just have a bunch of links hinting to comments added after its establishment. Assuming the people implementing the utilities wrote the manual pages, I suppose blame could fall upon either Joy or McKusick (Joy designed the Berkeley FFS, McKusick implemented it over the summer of 81 or 82), maybe the former since it endured through at least most of SunOS 4.x. It would explain much about it being reinserted into the manpage in SunOS 4.1.1 if Bill Joy was the one who originally wrote it. =)

All major BSD vendors still maintain this trailing sentence in the BUGS section.

And Peter Salus mentions nary a word about it.

  • Post a new comment


    default userpic