On Wednesday 22 June 2011 11:43:20 Neil Bothwick did opine thusly: > On Wed, 22 Jun 2011 11:53:19 +0200, Alan McKinnon wrote: > > > As Daniel said, this is what portage needs, it's been that > > > way since the autounmask stuff was introduced. > > > > In other words, we must now all end up with giganticly bloated > > package.use files to satisfy every "built with use" requirement > > system-wide? What's wrong with looking at the defaults and > > saying > > "Gee, you know what, the implicit rules on the box are going to > > do the right thing anyway, so let's proceed and build the > > stuff?" > > Are you saying the fltk is/would be built with -cairo anyway and the > recommended addition changes nothing? That sounds extremely > undesirable as it would make maintaining package.use much harder.
Ah, hang on a sec. It's not quite what I thought. The original emerge command done again, plus just fltk on it's own: # USE="dillo" emerge -pv claws-mail ... Calculating dependencies... done! [ebuild N ] net-libs/libetpan-1.0 USE="berkdb gnutls sasl ssl - debug -ipv6 -liblockfile" 0 kB [ebuild N ] x11-libs/fltk-2.0_pre6970-r1 USE="jpeg opengl png xft xinerama zlib -cairo -debug -doc" 0 kB [ebuild N ] www-client/dillo-2.2 USE="gif jpeg png ssl -doc - ipv6" 0 kB [ebuild R ] mail-client/claws-mail-3.7.9-r1 USE="crypt dbus dillo* gnutls imap ldap session spell ssl startup-notification - bogofilter -doc -ipv6 -nntp -pda -smime -spamassassin -xface" 0 kB # emerge -pv fltk ... Calculating dependencies... done! [ebuild N ] x11-libs/fltk-2.0_pre6970-r1 USE="cairo jpeg opengl png xft xinerama zlib -debug -doc" 0 kB The first and second are very different. > I was under the impression portage only did this if the USE flags > for a package needed to be changed from the current settings. Now it appears that emerge output (at least in the case of an unsatisfied emerge) is "what portage *needs* to do" instead of "what it *will* do" Portage has always displayed the latter right? That makes sense - you can see what the emerge command would do as entered and compare it to the error to see what the problem is. In this case it's a tweak to package.use which I'm perfectly happy to do. I think it's bug time, portage is displaying the wrong output for failures. -- alan dot mckinnon at gmail dot com
