[Yum-devel] [PATCH] yum/yumRepo does not uses URLGrabber provided by yum extension

Zdenek Pavlas zpavlas at redhat.com
Thu Sep 8 07:31:36 UTC 2011


> + def _createDefaultGrabber(**kwargs): 
> +     ug = self._getgrab()

You may as well use self.grab directly.

> +     if ug == nil:
> +         return URLGrabber(kwargs)
> +     return ug

None instead of 'nil', perhaps?
But _getgrab() may never return that.

> def _getMetalink(self): 
> ...
> + ug = _createDefaultGrabber(bandwidth = self.bandwidth,

Are you sure this does not trigger an infinite recursion?

_getMetalink() => _createDefaultGrabber() => _getgrab() 
=> _setupGrab() => self.urls => _geturls() => _baseurlSetup()
=> self.metalink_data => _getMetalink()

grabber self.grab is mirrors-aware, so you can't use it 
to fetch the mirror list itself.  There's a self.grabfunc 
that does not use mirror lists (and is one of the many 
write-only attributes in yum), but _setupGrab() sets both
at the same time so can't use it either.  

That's likely why getMetalink() instantiates it's own grabber.
What's the issue with that?  Does the yum-s3 plugin replace
self._grab with a custom grabber?  Does it so before
_getMetalink(), so it should use the one that's there already?
It seems likely, because otherwise it'd fail.

Then it'd probably make sense to do something like:

    ug = self._grab
    if not ug:
        # no grabber yet, make our own
        ugopts = self._default_grabopts()
        ug = URLGrabber(..., **ugopts)


More information about the Yum-devel mailing list