* Rusty Ballinger (rusty@xxxxxxxxxxxxxxxxxx) [001009 19:09]:
> > A surprise:
> > Rhino seems to want fam.h to build out the FileBasedCategory object.
> >
> > Some resulting questions:
> > Will rhino continue to use fam on linux instead of directly using
> > notify locks?
>
> Yes, as long as people working on rhino don't know anything about notify
> locks. :) You have a pointer to more info?
I went looking.. It's a 2.4 thing and rather new, and I can't find docs,
and.. I think you probably already know more about this than I do, reading
your prior participation in Directory notification threads.
I think I missed the point of fam, in that it is supposed to work even
with remote filesystems where NFS (and similar) cannot be forced to
respect leases, AFAIK.
Anyway, I was referring to the FL_LEASE lock type which went in with
2.4.0-test9-pre7 which should allow file watching to occur without the
imon patch.
I feel though, that it should be possible to cause fam calls to fall
through to direct watches on the local system via lease locks in the case
where the files are not network-provided. Helper daemons for filesystem
operations aren't something I want more of... I'm a newbie on this topic
though.
> > How terrible a thing would it be to build rhino without the
> > FileBasedCategory? Rhino seems not to depend further on it. Do most
> > other programs?
>
> No, the only thing which uses it is the example category which is built by
> mkrhinoism. You should be able to comment it out of the build with no
> problems. (In addition to lib/Makefile, you'll have to remove it from
> build/rpm/rpm.spec if you're building RPM packages.)
Yes, this is the approach I'm taking. Ideally I think it should be
possible to say ./configure --no-fam . Perhaps I should be the one to
implement this after I learn automake and friends.
> The other alternative is to install the fam header & library. If you're
> installing it from RPM, make sure you install --noscripts or whatever,
> because the install scripts want to add fam to your inetd.conf.
This is okay, the build system constructs a temporary distribution and
then throws it away, so any collateral damage is not important. However,
as I'm not planning on making a fam package right now, or requireing it
for failsafe which in turn doesn't use it, I'm not so interested in taking
this path.
> --Rusty
-josh
--
_,,wwww,,_
,wØØØ^"^ØØØØØØw,
.dØØØØI,_,Ø***ØØØØb. _____ _____ ______ ____
.ØØØ^ØØ*^'¸ ,¯'ØØØØØ. / ___/__ __/ ___// ____/ / _/___ _____
:ØØØbwww** `7,'ØØØØ: \__ \/ / / /\__ \/ __/ / // __ \/ ___/
`ØØØØØØØ7' dØØØØØØ' ___/ / /_/ /___/ / /___ _/ // / / / /___
`7ØØØØØ' ,* ØØØØØ7' /____/\__,_//____/_____/ /___/_/ /_/\___(_)
`*Øbb. "Ø, ØØØ*'
¯'^^ ^' '¯
--
Source code, list archive, and docs: http://oss.sgi.com/projects/rhino/
To unsubscribe: echo unsubscribe rhino | mail majordomo@xxxxxxxxxxx
|