On Thursday 31 May 2007 04:37:52 Boyd Stephen Smith Jr. wrote: > If it's really useful, it should be chmod'd +x and installed or linked > into /usr/sbin. I can do that myself (in that case /usr/local/sbin), but > such wonderful utilities shouldn't be hidden. :)
I can't really disagree with that, heh. It has been added to the faq though, however, it won't go online till the next release. > > It does still lack a couple of features though. > > Namely --library and the ability to pick another version when the ebuild > > for an installed version has been removed. At least the latter is soon to > > come. > > I noticed it was also trying to reinstall all my binary-only packages: > sun-jdk, blackdown-jdk, skype, and emul-linux-x86-compat. I guess > revdep-rebuild has some blacklist that prevents it from doing so. Ah yeah, forgot that one. It doesn't parse the variables that control where revdep-rebuild should and shouldn't look and what libs it should ignore. Instead it iterates over each installed package and runs ldd on every *.la and *.so* file. > > But at least it will never use the horrible hack that revdep-rebuild uses > > with --package-names because of the lack of support for the > > =category/package-version:slot syntax in portage-2.0* which it still > > supports.. :) > > You mean that portage actually supports slot deps now? Yes. There shouldn't have been an equal sign though. Using it in the tree is against policy, however, until we get an EAPI bump. But the new python-updater does take advantage of it and drops the horrible hacks it used to use. Also because portage has become much better at figuring out the build order with --deep when handed a list of broken packages in random order. Really dropping the horrible hacks in revdep-rebuild would be easy when just raising the dependency on portage to >=2.1.2. -- Bo Andresen
signature.asc
Description: This is a digitally signed message part.
