From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Petazzoni Date: Sun, 17 Jun 2018 22:48:37 +0200 Subject: [Buildroot] [PATCH 0/9] qt5: install missing examples In-Reply-To: <20180617200551.l2kb6yky7w3mfacg@archlinux> References: <20180612123206.31523-1-gael.portay@savoirfairelinux.com> <20180617144442.3fc03f6e@windsurf> <20180617200551.l2kb6yky7w3mfacg@archlinux> Message-ID: <20180617224837.33949a3a@windsurf> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hello Ga?l, On Sun, 17 Jun 2018 16:05:51 -0400, Ga?l PORTAY wrote: > > However, I am not super happy with how we install Qt5 packages right > > now. What we do is: > > > > I am not happy too; and I was already thinking about improving that. Great. > > - Staging installation: "make install", with no DESTDIR or anything > > similar > > > > - Target installation: manual copy. > > > > Ideally, I'd prefer if we could do: > > > > - Staging installation: make DESTDIR=$(STAGING_DIR) install > > - Target installation: make DESTDIR=$(TARGET_DIR) install > > > > Unfortunately, this doesn't work as-is, because the $(STAGING_DIR) > > location is hardcoded at Qt configure time, so it doesn't work > > immediately to do separate installation for staging and target. I know > > Arnout briefly looked into this during the last Buildroot hackathon. > > There is INSTALL_ROOT variable which looks to be equivalent to DESTDIR. Yes, but no. The problem is that with the way things are currently setup, if you set INSTALL_ROOT to /foo/bar and STAGING_DIR is /home/thomas/buildroot/output/staging, then Qt5 stuff will be installed in: /foo/bar/home/thomas/buildroot/output/staging Which is obviously wrong. And I believe the fact that it does this is due to: -sysroot $(STAGING_DIR) \ when configuring Qt5. Some research is needed to see if a better solution is possible :-) > Also, I was thinking about a qmake-package or a qt5module-package. A qmake-package infrastructure might make sense. But beware of Qt4/Qt5 compatibility. But to me, fixing the first problem is much more important/useful. Thanks! Thomas -- Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons) Embedded Linux and Kernel engineering https://bootlin.com