On Thu, Apr 08, 2004 at 07:46:48PM +1000, matthew green wrote:
>   On Thu, 8 Apr 2004, Jaromir Dolecek wrote:
>   > Why are the descriptions compiled into kernel? That seems as a waste
>   > of space. They are perfectly fine being in the manpage, IMHO.
>   
>   Given that many sysctls are undocumented, there seems to be a certain use
>   for putting the description near the code. Also, updating the kernel won't
>   update the manpage.
>
>there is a fine place for documentation and it is *not* in the kernel
>text/data.  unless there is a compelling reason for this change (and
>"keeping sysctl documentation upto date" doesn't even begin to count),
>i don't see that it's a good one in the slightest.

like i said in the commit message (which has been lost), if you don't
want them, you don't have to have them.  as for me, the 152 (out of
the 425 nodes in my tree) that are done cost me about 5k, but i like
knowing what all that stuff is, without having to wade through who
knows which man page?

       text    data     bss     dec     hex filename
    3436499   88004  281648 3806151  3a13c7 /NetBSD/THAT#308
    3441912   88004  281648 3811564  3a28ec /NetBSD/THAT#310

some things aren't documented at all, and this at least provides a
*comment* about what it might be.  modulo the fact that lkms and
devices can add their own nodes, upgrading your kernel doesn't update
your man page, and this makes the information *so* much easier to
find.  at the very least, these strings could be coopted to refer to
man pages.  i don't know which man page describes what
net.inet6.ip6.use_deprecated does; sysctl.8 merely makes mention of
its existence.

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior@daemon.org             * "ah!  i see you have the internet
twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
werdna@squooshy.com       * "information is power -- share the wealth."