On 21 May 2017 at 19:59, Ken Moffat <[email protected]> wrote: > On Sun, May 21, 2017 at 07:14:22PM +0100, Richard Melville wrote: > > On 21 May 2017 at 16:35, Ken Moffat <[email protected]> wrote: > > > > > > > > I disagree. My logs from last week show that package creating that > > > directory. Which package do _you_ think created that directory ? > > > > > > > Ken, that's a very good question. Unfortunately, if a package builds > > without problems I don't keep the logs. I build everything in a separate > > build directory, and then clear that directory on success. Maybe I > should > > keep all the build logs in a separate directory, but I've seen no point > -- > > until now, maybe. > > > > All I can say is that, on this build, I've only just installed > > hicolor-icon-theme and /usr/share/icons/hicolor was already in existence, > > created by an earlier package. Clearly, I could track this down as I > have > > a list of all the packages I've installed, but it will take some time. > > > > The reason that I'm so certain is because when the hicolor-icons-theme > > build initially failed I checked /usr/share/icons and hicolor existed > with > > a number of entries. As I've been working on this build sporadically > over > > the last few months it was dated 18 March. Then, when the build of > > hicolor-icon-theme succeeded, the date of the directory changed to 21 May > > (today) as it was overwritten. Anyway, if I have the time I'll see if I > > can dig out the answer. > > > > Richard > > Richard, no need to do that. Just look at the files (png) within > /usr/share/icons/hicolor. I do not have many on this build (very > little gnome, no kde, very little xfce) and mine come from cups, > evince, gimp, gtk3, inkscape, libreoffice, parole, qupzilla, vlc, > xfburn and xine. > > I guess some packages will install their icons even if the directory > is not present, but without hicolor-icon-theme itself the index > presumably cannot be created. > > I'm building hicolor-icon-theme shortly after I build > gobject-introspection and dbus-glib in the expectation that it > should be available for anything that wants to use it. > > So, for your build-order you are correct that h-i-t did not create > that directory, because it already existed. >
I was just in the process of replying to you. Finding the answer was much easier that I imagined; I forgot that I was tracking all my installs with porg. The answer is that both djvulibre-3.5.27 and emacs-25.1 previously created, and/or wrote to, /usr/share/icons/hicolor before the hicolor-icon-theme install. Richard
-- http://lists.linuxfromscratch.org/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
