On Nov 26, 2007 4:43 AM,  <[EMAIL PROTECTED]> wrote:
> As implementation note I would suggest that the MANPATH is searched while
> it is constructed, not constructed first completely and then searched.

I can take a look at that approach to see how involved it gets.  I'll
get back to you on this.

> To eliminate duplicates, realpath() isn't really nessary; just caching the
> results of stat() on the directory should be sufficient.

I had considered that as well.  However, you advised[1] the use of
realpath.  As an afterthought using realpath() also has the side
effect of making the output of man -l more user-friendly as well as
making error messages that complain about windex missing more
sensible.

1. 
http://mail.opensolaris.org/pipermail/opensolaris-code/2007-November/006393.html

> One additional consideration is potentially the behaviour of
>
> PATH=/usr/ucb:/usr/bin
>
> vs
>
> PATH=/usr/bin:/usr/ucb
>
> I'd like to get the proper "ps" manual page in both cases.

Now we are trampling on man.cf, which does tend to exist.

> Come to think of it, it would also be nice to be able to say:
>
>         man /usr/bin/ps

I've long wanted this.  Aside from the /usr/bin vs. /usr/ucb issue,
this would be very straightforward to do.

> and have this expand to something akin to:
>
>         man -M /usr/bin/../share/man ps
>
> (again, tricky for /usr/ucb which needs to be special cased, if we actually
> care about that as the man behaviour is currently broken in that case
> anyway)

This would involve possibly overriding the behavior configured by
man.cf.  Since *Solaris ships with /usr/share/man/man.cf adding this
functionality only in the absence of man.cf would be pointless.
Overriding the behavior configured by man.cf just because MANPATH is
not set seems like a confusing configuration option.

When taking it to this granularity, there are other things to consider
at the same time.  For instance, Should it also translate gnu info
pages into man pages and display them when appropriate?

I think that this work sounds like good follow-on work.  If others
feel strongly that /usr/ucb and/or gnu info handling belongs with this
and is unlikely to derail the ARC case I could look to lump it in.

-- 
Mike Gerdts
http://mgerdts.blogspot.com/
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to