On 09/26/2011 04:36 PM, Richard Purdie wrote:
On Mon, 2011-09-26 at 16:30 +0400, Dmitry Eremin-Solenikov wrote:
On 09/25/2011 06:50 PM, Koen Kooi wrote:
Op 25 sep. 2011, om 13:33 heeft Richard Purdie het volgende geschreven:
I've been looking at the qt3 failures on the autobuilder. It seems there
is some kind of contamination between the qt4 builds and the qt3 builds
and I'm not sure what is or isn't expected to work there.
We're doing the following in the build:
a) Build core-image-lsb (includes qt4, builds qt4+qt4-native)
b) Wipe out tmp
c) Add in meta-qt3 layer
d) Enable sstate from build in a)
e) Build core-image-lsb-qt3 which adds in qt3
f) Bitbake determines it needs to install qt4 for the image but
nothing depends on the native qt4 tools so these are not installed
g) This will build qt3+qt3-native with qt4 in the sysroot but no
qt4-native
h) We see build failures as qt3 seems to want to find bits of qt4-native
If qt4 and qt4-native are present, qt3 works
If qt4 and qt4-native are not present, qt3 works
If qt4 is present but qt4-native is not, qt3 fails
How do we fix this?
I suspect qt3 is missing bits of the tool renaming we do for qt4 and qt2 (moc,
uic, lrelease, etc)
Most probably it's true. However for qt3 it can be harder, as qt4
already provides uic3, etc.
If there is overlap, the files could be installed into subdirectories
under bin and then inserted into PATH by the appropriate .bbclass
files...
For qt2/opie we just install uic/moc as uic2/moc2 and the select
appropriate tools by passing OE_xxx_MOC (I don't remember xxx ATM)
environment variable to qmake/make.
--
With best wishes
Dmitry
_______________________________________________
Openembedded-core mailing list
Openembedded-core@lists.openembedded.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
