Bill Sommerfeld wrote:
> On Tue, 2006-10-10 at 10:22 -0700, Eric Lowe wrote:
>>        o An optional unit character, from the following set:
> ....
> I was a little surprised that the final unit character is optional since
> it's not optional for most printf-like functions.  What's the default
> behavior if the unit character is omitted?  Can you provide examples
> showing the behavior when it's omitted?

This was a mistake in the spec, the unit character was mandatory in the 
original spec from Yann but I munged it and it got missed in the final 
review. I'll fix it.

> I'm also a little surprised that there isn't an unsigned variant.
> Existing practice (for instance in "usr/src/cmd/ls/ls.c") has the
> equivalent function taking an unsigned value; types such as fsblkcnt_t
> are also unsigned.

I don't see any problem with adding an unsigned variant also. Yann?

> Will this project involve the conversion of existing utilities (ls, df,
> zfs, zpool, etc.,) within OS/Net to use the new common humanize_number
> function?

Since Yann is the contributor for this change, I'll let him answer this. :)

- Eric


Reply via email to