Re: head -n -1

[Paul Eggert]

address@hidden (Paul Jarc) writes:

> "tail -n +3" is the one that's different.

Yes, and for symmetry I suppose "head -n +3" could be different too.

However, I'd prefer it if someone would clear this with the POSIX
committee first.  The POSIX spec
<http://www.opengroup.org/onlinepubs/009695399/utilities/head.html>
says that -n's option-argument must be a "positive decimal integer",
and one could argue that "+3" is a positive decimal integer and that
"head -n +3" must therefore behave like "head -n 3" does.  (This is
what GNU "head" currently does.)

So, basically, the question you'd have to ask is, "Must a conforming
implementation treat 'head -n +3' the same way that it treats 'head -n
3'?".  But you'd have to give more justification for the question,
compare "head" to "tail", etc.

You can ask such questions formally here:

http://www.opengroup.org/austin/defectform.html

and you can see recent examples of such questions here, which you
can use for models:

http://www.opengroup.org/austin/mailarchives/ag-review/

If you'd like to draft the question I can help by proofreading it
before you submit it.