cvs commit: src/lib/libufs bread.3 cgread.3 getino.3 libufs.3 sbread.3 ufs_disk_close.3

[Jacques A. Vidrine]

On Thu, Oct 09, 2003 at 01:22:13PM -0700, Hiten Pandya wrote:
> On Thu, Oct 09, 2003 at 07:13:55AM -0500, Jacques A. Vidrine wrote:
> : I think you removed the wrong header.  Is <sys/param.h> required in
> : each of these cases?  If not, then I believe the synopsis should refer
> : only to <sys/types.h>:
> : 
> :   (1) Including <sys/types.h> is (by definition) less work than
> :       <sys/param.h>.
> 
> 	I remove the most logical one and also, there was some
> 	discussion about it on IRC.  Also, I was doing an mdoc
> 	review of all the lib manual pages, and I thought I would fix
> 	this, but I did not do a deep code review to see which header
> 	fits best for libufs, as I did not see a reason for it.
>
> :   (2) <sys/types.h> is a POSIX header, while <sys/param.h> is not.
> 
> 	LibUFS is not a POSIX library anyway, so I don't see what this
> 	has got to do with anything.  If anyone is inclined on me
> 	backing this out, I don't mind, although, I made a
> 	judgement/logical call when committing this.

Bruce pointed out that <sys/param.h> is actually needed, so there is no
issue.  You made the right call, I just couldn't tell from the commit.

On could argue that <sys/types.h> is needed for `size_t', regardless of
whether our implementation hides that fact, but I won't go there :-)

Regarding:
> 	LibUFS is not a POSIX library anyway, so I don't see what this
> 	has got to do with anything.

If there is a choice between a standard header and a system-specific
header, go for the standard header.  In this case, it wasn't a choice.

Cheers,
-- 
Jacques Vidrine   . NTT/Verio SME      . FreeBSD UNIX       . Heimdal
nectar at celabo.org . jvidrine at verio.net . nectar at freebsd.org . nectar at kth.se