Patch for BZ 303791?

[Clark Williams]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Below is a patch for the $SUBJECT bz. I originally just changed the case of the
search string to match what 'yum resolvedep' was printing, but then I thought
about
whether folks might not have the changed version of yum...

After I took an aspirin to deal with the dependency combinatorial explosion headache,
I looked at the code again. I didn't really want to pull in the re package just for
this one case, so I resorted to doing a case-insensitive match, which I *hope* will
deal with most cases.

What do you guys think?

Clark

diff --git a/mock.py b/mock.py
index 54b8f8c..35c3592 100644
- --- a/mock.py
+++ b/mock.py
@@ -355,9 +355,11 @@ class Root:
         # pass build reqs (as strings) to installer
         if arg_string != "":
             (retval, output) = self.yum('resolvedep %s' % arg_string)
+            searchstr = 'no package found for'
             for line in output.split('\n'):
- -                if line.find('No Package found for') != -1:
- -                    errorpkg = line.replace('No Package found for',
'')
+                idx = line.lower().find(searchstr)
+                if idx != -1:
+                    errorpkg = line[idx+len(searchstr):]
                     error(output)
                     raise BuildError, "Cannot find build req %s. Exiting." %
errorpkg
             # nothing made us exit, so we continue
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iD4DBQFHBSWXHyuj/+TTEp0RAtzCAJYy31hDn5zsGedHFM/LcWnrBbR5AKDnO10u
> ou94vu8Kf369iaF/VKiE7w==
> =iG5Y
> -----END PGP SIGNATURE-----

[Seth Vidal]

On Thu, 2007-10-04 at 12:40 -0500, Clark Williams wrote:
>  -----BEGIN PGP SIGNED MESSAGE-----
>  Hash: SHA1
>  
>  Below is a patch for the $SUBJECT bz. I originally just changed the case of the
>  search string to match what 'yum resolvedep' was printing, but then I thought
> about
>  whether folks might not have the changed version of yum...
>  
>  After I took an aspirin to deal with the dependency combinatorial explosion headache,
>  I looked at the code again. I didn't really want to pull in the re package just for
>  this one case, so I resorted to doing a case-insensitive match, which I *hope* will
>  deal with most cases.
>  
>  What do you guys think?
>  
>  Clark
>  
>  diff --git a/mock.py b/mock.py
>  index 54b8f8c..35c3592 100644
>  - --- a/mock.py
>  +++ b/mock.py
>  @@ -355,9 +355,11 @@ class Root:
>           # pass build reqs (as strings) to installer
>           if arg_string != "":
>               (retval, output) = self.yum('resolvedep %s' % arg_string)
>  +            searchstr = 'no package found for'
>               for line in output.split('\n'):
>  - -                if line.find('No Package found for') != -1:
>  - -                    errorpkg = line.replace('No Package found for',
> '')
>  +                idx = line.lower().find(searchstr)
>  +                if idx != -1:
>  +                    errorpkg = line[idx+len(searchstr):]
>                       error(output)
>                       raise BuildError, "Cannot find build req %s. Exiting." %
> errorpkg
>               # nothing made us exit, so we continue 
Is it time to start writing simple, custom yum-utils just for mock? So
it can call them as a program but also so mock can more rigidly control
its output?

-sv

[Clark Williams]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

seth vidal wrote:
>  On Thu, 2007-10-04 at 12:40 -0500, Clark Williams wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > Below is a patch for the $SUBJECT bz. I originally just changed the case of the
> > search string to match what 'yum resolvedep' was printing, but then I thought
> about
> > whether folks might not have the changed version of yum...
> >
> > After I took an aspirin to deal with the dependency combinatorial explosion
> headache,
> > I looked at the code again. I didn't really want to pull in the re package just
> for
> > this one case, so I resorted to doing a case-insensitive match, which I *hope* will
> > deal with most cases.
> >
> > What do you guys think?
> >
> > Clark
> >
> > diff --git a/mock.py b/mock.py
> > index 54b8f8c..35c3592 100644
> > - --- a/mock.py
> > +++ b/mock.py
> > @@ -355,9 +355,11 @@ class Root:
> >          # pass build reqs (as strings) to installer
> >          if arg_string != "":
> >              (retval, output) = self.yum('resolvedep %s' % arg_string)
> > +            searchstr = 'no package found for'
> >              for line in output.split('\n'):
> > - -                if line.find('No Package found for') != -1:
> > - -                    errorpkg = line.replace('No Package found for',
> '')
> > +                idx = line.lower().find(searchstr)
> > +                if idx != -1:
> > +                    errorpkg = line[idx+len(searchstr):]
> >                      error(output)
> >                      raise BuildError, "Cannot find build req %s. Exiting."
> % errorpkg
> >              # nothing made us exit, so we continue
>  
>  Is it time to start writing simple, custom yum-utils just for mock? So
>  it can call them as a program but also so mock can more rigidly control
>  its output?
>   
Maybe....

Are you talking about stuff that is a part of mock, a part of yum, or a separate
package altogether?

Clark

> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iD8DBQFHBSpJHyuj/+TTEp0RAmAJAKCTOzowhq8DMT71CrfRLON79c35HQCfYD1k
> S2qn/04z/xhtqRSgggvMd0o=
> =za13
> -----END PGP SIGNATURE-----

[Seth Vidal]

On Thu, 2007-10-04 at 13:00 -0500, Clark Williams wrote:
>  -----BEGIN PGP SIGNED MESSAGE-----
>  Hash: SHA1
>  
>  seth vidal wrote:
>  > On Thu, 2007-10-04 at 12:40 -0500, Clark Williams wrote:
>  >> -----BEGIN PGP SIGNED MESSAGE-----
>  >> Hash: SHA1
>  >>
>  >> Below is a patch for the $SUBJECT bz. I originally just changed the case of the
>  >> search string to match what 'yum resolvedep' was printing, but then I
> thought about
>  >> whether folks might not have the changed version of yum...
>  >>
>  >> After I took an aspirin to deal with the dependency combinatorial explosion
> headache,
>  >> I looked at the code again. I didn't really want to pull in the re package
> just for
>  >> this one case, so I resorted to doing a case-insensitive match, which I *hope*
> will
>  >> deal with most cases.
>  >>
>  >> What do you guys think?
>  >>
>  >> Clark
>  >>
>  >> diff --git a/mock.py b/mock.py
>  >> index 54b8f8c..35c3592 100644
>  >> - --- a/mock.py
>  >> +++ b/mock.py
>  >> @@ -355,9 +355,11 @@ class Root:
>  >>          # pass build reqs (as strings) to installer
>  >>          if arg_string != "":
>  >>              (retval, output) = self.yum('resolvedep %s' % arg_string)
>  >> +            searchstr = 'no package found for'
>  >>              for line in output.split('\n'):
>  >> - -                if line.find('No Package found for') != -1:
>  >> - -                    errorpkg = line.replace('No Package found for',
> '')
>  >> +                idx = line.lower().find(searchstr)
>  >> +                if idx != -1:
>  >> +                    errorpkg = line[idx+len(searchstr):]
>  >>                      error(output)
>  >>                      raise BuildError, "Cannot find build req %s.
> Exiting." % errorpkg
>  >>              # nothing made us exit, so we continue
>  > 
>  > Is it time to start writing simple, custom yum-utils just for mock? So
>  > it can call them as a program but also so mock can more rigidly control
>  > its output?
>  > 
>  
>  Maybe....
>  
>  Are you talking about stuff that is a part of mock, a part of yum, or a separate
>  package altogether? 
part of mock.

-sv

[Michael E Brown]

On Thu, Oct 04, 2007 at 02:06:12PM -0400, seth vidal wrote:
>  
>  On Thu, 2007-10-04 at 13:00 -0500, Clark Williams wrote:
>  > -----BEGIN PGP SIGNED MESSAGE-----
>  > Hash: SHA1
>  > 
>  > seth vidal wrote:
>  > > On Thu, 2007-10-04 at 12:40 -0500, Clark Williams wrote:
>  > >> -----BEGIN PGP SIGNED MESSAGE-----
>  > >> Hash: SHA1
>  > >>
>  > > Is it time to start writing simple, custom yum-utils just for mock? So
>  > > it can call them as a program but also so mock can more rigidly control
>  > > its output?
>  > > 
>  > 
>  > Maybe....
>  > 
>  > Are you talking about stuff that is a part of mock, a part of yum, or a separate
>  > package altogether?
>  
>  part of mock. 
I agree. More of a long-term thing, though. Here is my list of stuff we
need to work on:

1) "inside-out" setuid scheme (part of an overall revamp on how we see
security for mock running. We currently dont have a coherent strategy or
vision).
2) use yum-utils rather than calling cmdline
3) better caching (downloaded yum stuff, chroot cache)
4) more reliable mount/umount
5) Speedup options: ccache/distcc

--
Michael

[Clark Williams]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Michael E Brown wrote:
>  On Thu, Oct 04, 2007 at 02:06:12PM -0400, seth vidal wrote:
> > On Thu, 2007-10-04 at 13:00 -0500, Clark Williams wrote:
> >> -----BEGIN PGP SIGNED MESSAGE-----
> >> Hash: SHA1
> >>
> >> seth vidal wrote:
> >>> On Thu, 2007-10-04 at 12:40 -0500, Clark Williams wrote:
> >>>> -----BEGIN PGP SIGNED MESSAGE-----
> >>>> Hash: SHA1
> >>>>
> >>> Is it time to start writing simple, custom yum-utils just for mock? So
> >>> it can call them as a program but also so mock can more rigidly control
> >>> its output?
> >>>
> >> Maybe....
> >>
> >> Are you talking about stuff that is a part of mock, a part of yum, or a separate
> >> package altogether?
> > part of mock.
>  
>  I agree. More of a long-term thing, though. Here is my list of stuff we
>  need to work on:
>  
>  1) "inside-out" setuid scheme (part of an overall revamp on how we see
>  security for mock running. We currently dont have a coherent strategy or
>  vision).
>  2) use yum-utils rather than calling cmdline
>  3) better caching (downloaded yum stuff, chroot cache)
>  4) more reliable mount/umount
>  5) Speedup options: ccache/distcc 
All good. I agree completely.

Now tell me if my short-term fix for $SUBJECT is acceptable :)

Clark
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iD8DBQFHBS/vHyuj/+TTEp0RAqVWAJ4lK0QwHJ6Dda8VXQJD+q9ITga5hQCgkWP2
> OXs1pxSeTnM6lmoEiv3S2Ho=
> =l1bJ
> -----END PGP SIGNATURE-----