svn commit: r263110 - head/share/man/man4

John-Mark Gurney jmg at funkthat.com
Fri Mar 14 17:26:50 UTC 2014


Christian Brueffer wrote this message on Fri, Mar 14, 2014 at 17:55 +0100:
> On 3/14/14 5:47 PM, Warner Losh wrote:
> > 
> > On Mar 14, 2014, at 10:41 AM, John-Mark Gurney <jmg at funkthat.com> wrote:
> > 
> >> Warner Losh wrote this message on Fri, Mar 14, 2014 at 08:30 -0600:
> >>> On Mar 14, 2014, at 3:13 AM, Christian Brueffer <brueffer at FreeBSD.org> wrote:
> >>>
> >>>> On 3/14/14 2:54 AM, John-Mark Gurney wrote:
> >>>>> John Nielsen wrote this message on Thu, Mar 13, 2014 at 16:28 -0600:
> >>>>>> On Mar 13, 2014, at 10:19 AM, John-Mark Gurney <jmg at freebsd.org> wrote:
> >>>>>>
> >>>>>>> Author: jmg
> >>>>>>> Date: Thu Mar 13 16:19:36 2014
> >>>>>>> New Revision: 263110
> >>>>>>> URL: http://svnweb.freebsd.org/changeset/base/263110
> >>>>>>>
> >>>>>>> Log:
> >>>>>>> remove link to the missing AMD Geode LX SB man page... we can add it
> >>>>>>> back once someone cares enough to write one..
> >>>>>>
> >>>>>> You mean like this one?
> >>>>>> http://svnweb.freebsd.org/base/head/share/man/man4/man4.i386/glxsb.4
> >>>>>
> >>>>> The problems of checking on an amd64 box... :(
> >>>>>
> >>>>> Actually, how are we suppose to handle links to arch dependant man
> >>>>> pages in arch independant man pages?  I did this check on an amd64 box,
> >>>>> so the page glxsb didn't get installed...  Should we just always
> >>>>> install these man pages on all arches then?  Or are we fine w/
> >>>>> references to non-existant pages (on some arches)?
> >>>>>
> >>>>> Or should glxsb.4 be moved to an arch independant dir?
> >>>>>
> >>>>
> >>>> I wonder if it makes sense to keep arch-dependent man directories at
> >>>> all.  I.e., if I prepare an ARM disk image on my amd64 laptop, I really
> >>>> want to have the arm manpages available on that amd64 box.
> >>>>
> >>>> Does anyone see a reason not to move man4/man4.${ARCH}/*.4 to man4?
> >>>> We're talking about 60 manpages here, so space is not really an issue.
> >>>
> >>> Historically there was the separation because xp on one platform might be
> >>> radically different from xp on another platform. The other reason was confusion
> >>> because it wasn?t always clear if device foo actually worked on platform X.
> >>
> >> We have less of that issue now, but it could still be.. but issues like
> >> that can be addressed other ways, though kernel config, release notes,
> >> etc..   I doubt people use manpages as a guage if a device worked on
> >> their arch...  Plus, if the device is shared and known not to work on
> >> a specific arch, that should be listed in the BUGS section of the man
> >> page.. :)
> > 
> > Yea, kinda my point :)
> > 
> >>> Do we have any collisions like that? If so, we need to resolve those first.
> >>
> >> Doesn't look like it...
> >> $ grep ^_ Makefile  | sed -e 's/.4.*//' | wc -l
> >>      86
> >> $ grep ^_ Makefile  | sed -e 's/.4.*//' | sort -u | wc -l
> >>      86
> >>
> >> It's also telling that besides i386 and amd64 specific man pages, we
> >> only have one mips specific man page in man4?
> > 
> > True enough, I hadn?t looked...
> > 
> >> This won't address the other man pages for arch specific utilities
> >> and libraries though, like sunlabel?
> > 
> > True, but those are niche things, so it is less important :)
> > 
> >>> Also, it would be good to tag the ones that are arm specific as such somehow.
> >>
> >> Should we just tag the pages w/ a section listing relevant architectures,
> >> or break the beauty of single digit section numbers and do 4.i386?
> > 
> > Tag in the man page itself, either in its own unique section, or in BUGS.
> > 
> 
> We kinda maintain some of this information already (albeit I'm not sure
> how accurate it is) in the hardware notes.  The hardware notes
> themselves are generated from the HARDWARE sections of the manpages,
> while the arch information is in
> src/release/doc/share/misc/dev.archlist.txt.
> 
> One possibility would be to generate this file from an ARCHITECTURE (or
> the like) section in section 4 manpages, the same way as we to with
> HARDWARE sections.

This becomes a little troublesome to handle all the corner cases that
can be expressed, and what do we want to say...  There is if the code
has been tested or not, and just how arch specific it is...

Pretty much all of our supported platforms support PCI, and therefor
ISA through a bridge, which means all the "i386" specific ISA cards
could possibly find themselves on non-i386 machines...  Now if they'll
work due to endian issues or other issues or not is another issue
entirely...

Some of the entries in dev.archlist.txt don't match what you can build
per sys/conf/files*...  aic's pccard is available to all, it's isa is
only to i386 and pc98, but listed as i386, pc98 and amd64... so we end
up w/ an odd set here, that isn't very easy to figure out and parse...

Do we want a list of expected working and tested working?  Then we also
might want to list known not working configurations too...

I know too much about how to hack to get crazy hardware to work where
it isn't suppose to, so other people w/ clearer heads on this topic
should make these decisions...

-- 
  John-Mark Gurney				Voice: +1 415 225 5579

     "All that I will do, has been done, All that I have, has not."


More information about the svn-src-head mailing list