Geoffrey Young wrote:
I just tested and it seems to work fine for me under 'root'. Notice that
all files under t are chowned to the user/group the server is running
under (not-root) before the tests are run.


yes, for me too, but on linux. did you test any OSX variant, though?

richard, are you on darwin or panther?  I have access to darwin but not
panther, so my debugging is limited.

Richard is on darwin, please see the original report: http://marc.theaimsgroup.com/?l=apache-modperl-dev&m=106989642826330&w=2

I think I remember about a similar problem discussed on p5p

   # testing : $r->finfo->user()
   # expected: 4294967294
   # received: -2
   not ok 7
   # testing : $r->finfo->group()
   # expected: 4294967294
   # received: -2

Notice that the expected values are not very good.


but they match what Apache-Test is doing. from richard's output

** root mode: changing the files ownership to 'nobody' (4294967294:4294967294)

so perl is being consistent with itself, at least.  apr  merely reports back
the result from an fstat call, so the issue really does seem to be with the
underlying OS.

Cool, I've missed that. It just looked weird. I guess they use 2*32-1 for nobody, so they can accomodate 2*32-2 users ;)


Here it is:
http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-01/msg01530.html


those are good links.  however, according to nick in the first thread, they
all seem to be talking about c/mtime, not uid/gid.

if you look at that perl test, they don't test uid/gid at all. So I won't be surprised if that's the same case.


BTW, why Finfo doesn't use uid/gid but user/group? because of the C data struct names? It's probably more intuitive to rename them to be uid/gid, since they return the ids and not names.

and we also need to document that that finfo doesn't give a valid
user/group under Darwin/UFS.


I think that's premature at this point.  we can just keep skipping failing
platforms, I suppose, but I'd rather find someone who really understands
this OSX biz before we simply proclaim that the results are meaningless on
that platform.  that it works fine as non-root (with sensical 501:0 values)
leads me to believe that something else is going on.

sure, I suggested to do that if that's indeed the problem.


Geoff, you are taking care of this?


haven't I been?

I thought you bailed out, saying that we should document that the tests are not to be run as root. Hence I was asking, so I won't step on your toes.


I'll piddle around on moof a bit, but underlying POSIX stuff (or whatever it
is) really isn't my strong point.

How does perl handles that on darwin/UFS? We should just do the same.


__________________________________________________________________
Stas Bekman            JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/     mod_perl Guide ---> http://perl.apache.org
mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org   http://ticketmaster.com


--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to