The package mozilla-firefox and mozilla-firefox-bin should block each other as packages such as epiphany depend upon mozilla-firefox to build, thus when the user has mozilla-firefox-bin installed, emerging epiphany results in mozilla-firefox also being installed. openoffice and openoffice-bin block each other, so I believe that firefox should do the same.
*** This bug has been marked as a duplicate of bug 160857 ***
Why not? What reason is there not to do this?
If you want a valid reason why this should be done:
Suppose a user installs mozilla-firefox-bin.
They then go on to install a package that pulls in mozilla-firefox thus ending up with mozilla-firefox as a dependency and not recorded in the world file.
The user would most likely end up now running this install as the command firefox refers to mozilla-firefox and firefox-bin refers to the mozilla-firefox-bin install.
Now suppose a critical bug appears in firefox that exposes a security issue.
A user not used to the gentoo portage system may then unwittingly run emerge -uav world, sees the update to mozilla-firefox-bin install it and assume everything is fine - but of course not because they haven't updated the mozilla-firefox package which they are in general running instead of the bin package.
Yes you may argue that the user should become more used to the portage system, but if you want to see the spread of linux then making software installation and updating idiot-proof is required and this is one of those areas where things need to be improved.
(In reply to comment #2)
> Why not? What reason is there not to do this?
amd64 users may prefer a 64-bit firefox over the 32-bit firefox-bin, except when viewing flash/32-bit media.
There should be a virtual satisfied by either firefox. In my mind, there should be a virtual for each -bin package. But there's no reason to specify blockers.
(In reply to comment #4)
> (In reply to comment #2)
> > Why not? What reason is there not to do this?
> amd64 users may prefer a 64-bit firefox over the 32-bit firefox-bin, except
> when viewing flash/32-bit media.
> There should be a virtual satisfied by either firefox. In my mind, there
> should be a virtual for each -bin package. But there's no reason to specify
The problem with a virtual though is that the firefox-bin package doesn't provide the necessary headers for compiling against. It appears to me that this case needs a support mechanism at the level of portage to fulfill all requirements.
Please, stop bugging us with nonsense.