Age | Commit message (Collapse) | Author | Files |
|
make desktop errors fatal
|
|
|
|
|
|
With diverting packages ${PN} can sometimes not be used. In such cases
we don't to use the filename of the currently included file. This can
work with immediate assignments and a helper function like this one.
|
|
Experiment with having the same source build with different versions of
the underlying toolkit. In the best case we want to be installable in parallel.
|
|
qtopia
Change the include and library path to end with qtopia as this is where
qtopia-core is going to stage the headers in. The full blown Qtopia platform
will use the same directory.
|
|
(rcc)
The build of qsvn failed due invoking the wrong rcc. Change the qmake specs
to set QMAKE_RCC and set it to OE_QMAKE_RCC.
|
|
inherit qmake will download Qt3 and build parts of it, we don't want that.
|
|
|
|
and 'fbf7570478d5954f66dbaa425a8b6b1c8c1b862c'
|
|
|
|
did change
|
|
and '8f4f4a96be49747ae2c3c51e51854655b4fc65b6'
|
|
failure.
|
|
them due QA issues
Disable packaging of the examples/demos as we need to get rid off
the .debug directories inside the nested hierachy of the examples
and demos.
|
|
|
|
and 'c0e108d768a455005e5b9ad28796112f203c3de6'
|
|
cause no change to compilation or packaged content - if it built for you previously, then you don't need the patch, but it will do no harm.
|
|
|
|
|
|
* This code is sick.
|
|
default module
-Do not hardcode O2, we have flags in OE to do such things and passing both
didn't make sense
-With QtopiaCore QtGui is using Tcp/Udp/Sockets for the QWS implementation. This means when
building a gui application you will need the network module. Allow to add modules using
environment variables.
|
|
* Don't try to interpret APM status when it's not available.
* Don't assume X mode is always active.
|
|
* Fixes #3471.
|
|
decide the tuning!
|
|
between Embedded and X11
-Use the CROSSARCH hack/patch on QtopiaCore as well
-Do not search /usr/include for freetype and openssl on Qt as well
-Do not build a qmake version (needed a diversion due different configure scripts)
|
|
* Need to fix this, need to use good common codebase, because debugging
why some adhoc code doesn't work here or there is big waste of time.
|
|
* If there's something to unbreak... Do following:
1. Support logging. Software without logging is now software,
but a random wreck.
2. Don't do off-by-one errors.
3. Don't bug on system every 2 seconds. Instead, do a select
on X connection to catch touchscreen events ASAP. This doesn't
work for keys though, there's larger latency. As for other checks,
do them once in 5s.
4. The whole thing must die. Soon.
* Mark configs propery for ipkg.
* Should fix #3540.
|
|
and 'dafddd1d044d96ecf2e968c17c56e4f9f332240d'
|
|
the glob
-${PN}-doc already has the html documentation, no need for -documentation package
-Package the demos and exmaples
-.debug and translations are missing
|
|
python function to generate the files for the dev package
|
|
.so, .la, .prl in the long form
|
|
-Don't install every library to libQtUiTools-dev :}
-Try to put the .la and .prl into the -dev packages as well
|
|
-QT_LIBRARY_NAME lost the lib prefix to make it usable with .pc files
-Add headers to the -dev packages
-Look in the right directory for the plugins remains
-Get rid off the mkspecs before packaging
-Attempt to package translations
|
|
The time spent on recompiling outweights the time saved using pch. So disable
pch and have better recompile times.
|
|
buildsystem should handle deps
|
|
|
|
(bare).
|
|
|
|
|
|
matchbox-panel-manager.
* It seems this thing is standalone misfit which is supposed to be simply run
during session startup ;-\. Try to add at least bit of user configurability. With
this desktop file, it is at least possible to add it via context menu on
matchbox-panel. But it doesn't seem to get recorded in panel settings, so
cannot be removed via menu and its unclear if it will be restarted after
reboot ;-(.
|
|
|
|
NFS boot.
|
|
|
|
and 'd29274157cd7d80d66aa37ae4d3ebf85231e594f'
|
|
USB gadgets.
* Now that there's flexible early userspace support in OE (initramfs-uniboot)
which handles NFS booting among other things, there's no longer need to have
NFS boot support in kernel. This gets NFS out of kernel, and frees g_ether
from being compiled into kernel, thus opening road for modular USB gadget
drivers.
* A bit tough move deep into Angstrom RCs, but this was on TODO for long time,
and just yesterday I hit usecase where I was stuck late night with the need
to move data between 2 boxes, with only common interface in both being USB,
but still w/o flashdrive to use. So, I took this opportinity not just
to try g_file_storage, but also handle this issue.
|
|
and 'a2db4317a6940d3a7ea74bc07545523686942a59'
|
|
|
|
NFS root.
|
|
* Provides best practices for exports locations and sample of exports file.
(The same location is used by default by initramfs-uniboot).
* Tool to deploy tar.bz2/tar.gz/cpio.gz images to exported location.
|