[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

OT: manpage fun facts, or how much documentation *is* there anyway?

Background:  so I'm explaining to dear old Mom, I mean, dear Mom, how
there's tons of documentation available on GNU/Linux systems, that it's
typeset for the user on the fly through the man system, and how in fact
one of the first uses of UNIX[tm] by AT&T was as a document
editing/processing system through roff/nroff, and kin.

I've also long been a fan of 'man -Tps <manpage> | mpage -2 >
/tmp/<manpage>.ps; gv -seascape /tmp/<manpage>.ps' as well.  Try that if
you don't already know it.

So I'm shilling GNU/Linux to other people and wanted to make a point of
how much documentation there is available for it.  But getting a hard
page count is, well, hard.   So how about we just stick to manpages.

   - They all live in /usr/share/man/man* and /usr/X11/man/man*

   - We can get a pagecount by using standard text output and running
     through 'pr', which paginates the output (per your current
     papersize).  Grep for '      Page [0-9][0-9]* *$', and pipe through
     'tail -1' to get the total count for that manual.

   - Burn some cycles doing this on your system.

For my desktop, with 1631 packages installed (of 14,000+ in Debian
unstable), there are a total of 6,233 manpages, totalling 27,160 pages
of output, or about 4.3574 pages each.  apt-file tells me (after
unduplicating for multiple listings of packages among Debian releases)
that there are 60,013 manpages in the distribution total.  

Assuming the average pagecount holds, that's 261,500 pages of output.
My typical O'Reilly book runs about 500 pages, and I can fit about 30 to
a shelf, and five shelves to a bookcase, so that's about 3.5 bookcases
full of manual pages.

If you took this to Kinko's to be printed 2-up at 7c per page, that's
$9,152 dollars in docs, printing costs alone.  $18k if you've got lazy

...and we're not counting info pages, READMEs, HOWTOs, RFCs, GNU/Linux
Gazette, and the other documentation available on the system.  Which I
may total up later on.

From my own set of 6k+ manpages, the ten longest are:

  Rank  Pages
  ----  -----
     1  1118   ethereal-filter(4)
     2  335
     3  214    perltoc
     4  160
     5  151    smb.conf
     6  148    arm-palmos-gcc
     7  131
     8  118
     9  101
    10  81     bash

Note:  Because of the way I computed totals, I have to go back manually
and find the manpage in question, and haven't done so for all of the
above.  Several of the unnamed pages are various gcc versions' manuals.

Disclaimers:  this is text-formatted output, 80 column line, US 'letter'
format.  Postscript and text manpage output differs (PS is about 75% of
the pagecount of text), and other local variances may occur.

There's been some grousing in GNU/Linux circles about the quality of
documentation, particularly for a few high profile projects.  But darned
if we ain't got that quantity thing down.


Karsten M. Self <kmself@ix.netcom.com>        http://kmself.home.netcom.com/
 What Part of "Gestalt" don't you understand?
   GNU/Linux web browsing mini review:  Galeon v1.2.5.  Kicks ass.

Attachment: signature.asc
Description: Digital signature

Reply to: