Re: [PATCH] fix gprof's option parsing

[Bob Wilson <bwilson at tensilica dot com>]

On Oct 16, 2006, at 7:32 PM, Daniel Jacobowitz wrote:
> OK.  Could you fix the manual, too?  I believe it says that the
> arguments to -n and -N are optional.

Sure.  I hadn't thought to check that, but you are right.  I've  
included a patch below.  I'm inclined to leave the man page SYNOPSIS  
as it is:

gprof [ -[abcDhilLrsTvwxyz] ] [ -[ACeEfFJnNOpPqQZ][@var{name}] ]

It seems to imply that the argument for -n/-N is optional, but this  
syntax is already oversimplified.  (If you treat it literally as a  
regular expression, you could have "-name" with no option letter at  
all, and you could only have one of these options.)  What do you  
think?  An alternative would be to put:

[ -n@var{name} ] [ -N@var{name} ]

on a separate line -- that would be more accurate but makes the  
synopsis longer and less readable (I think).

2006-10-16  Bob Wilson  <bob.wilson@acm.org>

	* gprof.texi (Analysis Options): Argument for -n/-N is mandatory.

Attachment: gprof-doc.patch
Description: Binary data