Re: [QGIS-Developer] Build failure in master

[Tom Elwertowski]

Try deleting everything in the {build}/python/analysis folder. Just 
deleting the generated *.cpp files isn't enough.


I have been seeing a multiple definition sip_{module}part{n} 
sip_{module}part{n+1} error about once every three months for awhile. 
Deleting everything in {build}/python/{module} allows the build to be 
resumed (macOS Catalina).


Tom


On 7/8/21 4:59 PM, Tim Sutton wrote:
Thanks for testing Werner. A few others on Telegram also reported that 
things are building fine for them. I will try doing  clean build 
tomorrow, thanks!


Regards

Tim

On Thu, Jul 8, 2021 at 1:01 PM Werner Macho > wrote:


Hi Tim,

Just updated master and also tried to build with no errors whatsoever.
(Using ninja and ccache)

Maybe try a clean build?

regards
Werner

On Thu, Jul 8, 2021 at 1:10 PM Tim Sutton mailto:t...@kartoza.com>> wrote:

Hi All

Just updated to master and getting a build error:


Consolidate compiler generated dependencies of target
python_module_qgis__analysis
[ 68%] Linking CXX shared module ../output/python/qgis/_analysis.so
/usr/bin/ld:

CMakeFiles/python_module_qgis__analysis.dir/analysis/sip_analysispart13.cpp.o:/home/timlinux/dev/cpp/QGIS-QtCreator-Build/python/analysis/sip_analysispart13.cpp:138:
multiple definition of `sipTypeDef__analysis_QgsAnalysis';

CMakeFiles/python_module_qgis__analysis.dir/analysis/sip_analysispart12.cpp.o:/home/timlinux/dev/cpp/QGIS-QtCreator-Build/python/analysis/sip_analysispart12.cpp:1812:
first defined here
collect2: error: ld returned 1 exit status
gmake[2]: ***
[python/CMakeFiles/python_module_qgis__analysis.dir/build.make:1101:
output/python/qgis/_analysis.so] Error 1
gmake[1]: *** [CMakeFiles/Makefile2:6919:
python/CMakeFiles/python_module_qgis__analysis.dir/all] Error 2
gmake[1]: *** Waiting for unfinished jobs
Consolidate compiler generated dependencies of target qgis_gui
[ 78%] Built target qgis_gui
gmake: *** [Makefile:166: all] Error 2
12:03:00: The process "/usr/bin/cmake" exited with code 2.
Error while building/deploying project qgis (kit: Qt 5.15.2 (qt5))
When executing step "Build"


Is it because of the new exif alg maybe?

Regards

Tim


-- 
--


Tim Sutton
Visit http://kartoza.com  to find out about
open source:
  * Desktop GIS programming services
  * Geospatial web development
* GIS Training
* Consulting Services
Tim is a member of the QGIS Project Steering Committee

---
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org

List info:
https://lists.osgeo.org/mailman/listinfo/qgis-developer

Unsubscribe:
https://lists.osgeo.org/mailman/listinfo/qgis-developer




--
--
​

Tim Sutton
Visit http://kartoza.com  to find out about open 
source:

  * Desktop GIS programming services
  * Geospatial web development
* GIS Training
* Consulting Services
Tim is a member of the QGIS Project Steering Committee
---

___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Current state of QtWebKit dependency

[Tom Elwertowski]

QtWebKit 5.9.0 builds with Qt 5.13.0 on macOS 10.14.

Using Qt 5.12, the QMAKE_INCDIR_OPENGL path in 
clang_64/mkspecs/modules/qt_lib_gui_private.pri needs to be fixed. Qt 
5.11 and earlier are ok.


The latest QScintilla and PyQt5 also compile ok with Qt 5.13.0 and macOS 
10.14.



Tom


On 6/27/19 7:20 PM, Nyall Dawson wrote:

On Fri, 28 Jun 2019 at 05:27, Larry Shaffer  wrote:


What's the current state of QtWebKit dependency within the main source code?

I have seen this post and am curious if it is still true:
http://osgeo-org.1560.x6.nabble.com/QGIS-Developer-qgis2web-and-qtwebkit-tp5378226p5378445.html


Nothing's changed on either side, unfortunately. We still NEED
qtwebkit, and the situation upstream is still a confusing pile of
excrement and misinformation.


I'm assuming that if there is still a dependency on QtWebKit there is also a 
need for the PyQt bindings for its modules as well.


Yes. You can build without it, but that's really just a hack to allow
a build to proceed. The resulting QGIS app will be crippled and
missing core functionality.

Nyall



Regards,

Larry Shaffer
Dakota Cartography
Black Hills, South Dakota

___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] OpenCL and macOS

[Tom Elwertowski]

For macOS, copying cl2.hpp from 
https://github.com/KhronosGroup/OpenCL-CLHPP/releases to 
/usr/local/include/CL works.


Setting USE_OPENCL=OFF also works to disable OpenCL.

One or the other is necessary or macOS will try the use the v1.2 C 
headers provided by Apple and pointed to by OpenCL_INCLUDE_DIR.


After lanuching QGIS, I see the Options|Acceleration panel with three 
choices:

Intel(R) Core(TM) i7-3615QM CPU @ 2.30GHz
HD Graphics 4000
GeForce GT 650M

To automate the mac build, my suggestion is to create a 
OpenCL_HPP_INCLUDE_DIR variable. If a mac user sets it, use OpenCL; if 
not, don't use OpenCL. For other platforms, OpenCL_HPP_INCLUDE_DIR is 
automatically set to OpenCL_INCLUDE_DIR.


Tom

___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] OpenCL and macOS

[Tom Elwertowski]

The directory found by CMake 
(/System/Library/Frameworks/OpenCL.framework/Headers/) contains only .h 
files.


The khronos site mentioned by Denis offers cl.hpp for v1 and cl2.hpp for 
v2. qgsopenclutils.h includes cl2.hpp.


Should I download and try cl.hpp on my mac? If it compiles, how can I 
test running it.


Tom


On 8/9/18 10:12 AM, Denis Rouzaud wrote:

I think the issue is that there is no header installed on mac
https://stackoverflow.com/a/23079478/1548052

Le jeu. 9 août 2018 à 16:06, Alessandro Pasotti <mailto:apaso...@gmail.com>> a écrit :


On Thu, Aug 9, 2018 at 3:54 PM, Tom Elwertowski
mailto:telwertow...@comcast.net>> wrote:

Hi all,

A recent change added OpenCL. Compilation fails on macOS because
Apple provides v1.2 (macOS 10.13) while QGIS seems to require v2. 




That's wierd: 1.1 is what should be required can you file a ticket
and provide full logs?

Adding a version to FIND_PACKAGE will fix the macOS compile and
not use OpenCL.

Apple has deprecated OpenCL in favor of its own Metal
technology. OpenCL will remain for macOS 10.14 (Mojave, fall
2018) but may be removed after fall 2019.

For the future, either OpenCL must remain an optional QGIS
feature in order to support macOS or the API needs to be
abstracted so that macOS can use Metal while other OSs use OpenCL.



OpenCL is an optional feature and there is no plan to change that.



Tom
LinkedIn: https://www.linkedin.com/in/thomas-elwertowski-a0886032
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
<mailto:QGIS-Developer@lists.osgeo.org>
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer




-- 
Alessandro Pasotti

w3: www.itopen.it <http://www.itopen.it>
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org <mailto:QGIS-Developer@lists.osgeo.org>
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

--

Denis Rouzaud
de...@opengis.ch <mailto:de...@opengis.ch>
+41 76 370 21 22



___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

[QGIS-Developer] OpenCL and macOS

[Tom Elwertowski]

Hi all,

A recent change added OpenCL. Compilation fails on macOS because Apple 
provides v1.2 (macOS 10.13) while QGIS seems to require v2. Adding a 
version to FIND_PACKAGE will fix the macOS compile and not use OpenCL.


Apple has deprecated OpenCL in favor of its own Metal technology. OpenCL 
will remain for macOS 10.14 (Mojave, fall 2018) but may be removed after 
fall 2019.


For the future, either OpenCL must remain an optional QGIS feature in 
order to support macOS or the API needs to be abstracted so that macOS 
can use Metal while other OSs use OpenCL.


Tom
LinkedIn: https://www.linkedin.com/in/thomas-elwertowski-a0886032
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Suggestion for improved macOS build compatibility testing

[Tom Elwertowski]

I have posted my steps for building QGIS and dependencies from source 
here: https://gist.github.com/telwertowski/a047772c8ae38bb6612c0c7180e82a0c


Tom

On 4/28/18 7:12 PM, Jeremy Palmer wrote:


On Sun, Apr 29, 2018 at 5:23 AM, Tom Elwertowski 
<telwertow...@comcast.net <mailto:telwertow...@comcast.net>> wrote:


Currently, I install Qt binaries and Kyngchaos frameworks and use
configure/make for everything else. I can document the steps. I
haven't in the past because people asking for mac builds have said
they are not developers.

After review of my build process, we can determine if it can be
automated.

Big +1 for this and nightly builds :)

Cheers
Jeremy

___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Suggestion for improved macOS build compatibility testing

[Tom Elwertowski]

Hi Matthias,

I build QGIS dependencies mostly from source. I don't use homebrew or 
macports. When I started, fink was popular and very useful for the first 
few months. Once I became familiar with Unix-style tools however, 
building directly from source was less problematic than dealing with the 
external issues in package managers.


Currently, I install Qt binaries and Kyngchaos frameworks and use 
configure/make for everything else. I can document the steps. I haven't 
in the past because people asking for mac builds have said they are not 
developers.


After review of my build process, we can determine if it can be automated.

Tom


On 4/26/18 2:19 PM, Matthias Kuhn wrote:

Hi Tom,

Sorry to hear that. In the past we had macOS continuous integration 
tests in place but had to disable them, because the package manager used 
there (homebrew) was breaking dependencies unpredictably (I don't think 
that's unexpected news for you if you are building on mac though ;) )


There is no "verification machine" on linux, but there is a docker 
image. However I am not sure how easy it is to modify this image put 
headers in different folders and make all the executables (xxx-config) 
return the modified paths - while keeping the infrastructure maintainable.


Alternatively, I would very much like to get the mac builds back up and 
running. I don't think we can do this again with the upstream homebrew 
repo since we don't want our pull requests flagged erroneous for 
external problems. The best alternative I see is, that we need a curated 
list of formulae in a separate repository, which would not only help the 
testing but potentially also the release process. Would you be 
interested in taking a lead on this?


Best regards

Matthias


On 04/26/2018 07:58 PM, Tom Elwertowski wrote:

Hi all,

Both yesterday and today, the macOS build failed due to missing CMake 
INCLUDE specifications.


MacOS puts each framework in a separate directory while Linux merges 
everything into /usr/local. Thus, Linux commits will compile when a 
new dependency is added as long as some other dependency already 
specifies /usr/local.


Here is an idea that would help macOS build compatibility. Place each 
dependency on the verification build machine in a separate directory 
such as

/usr/local/gdal-2.2.3/
/usr/local/geos-3.6.1/

This would make commits fail upfront on Linux if CMake INCLUDES needed 
by macOS were missing.


Tom
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer



--
Matthias Kuhn
matth...@opengis.ch <mailto:matth...@opengis.ch>
+41 (0)76 435 67 63 <tel:+41764356763>
OPENGIS.ch Logo <http://www.opengis.ch>


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

[QGIS-Developer] Suggestion for improved macOS build compatibility testing

[Tom Elwertowski]

Hi all,

Both yesterday and today, the macOS build failed due to missing CMake 
INCLUDE specifications.


MacOS puts each framework in a separate directory while Linux merges 
everything into /usr/local. Thus, Linux commits will compile when a new 
dependency is added as long as some other dependency already specifies 
/usr/local.


Here is an idea that would help macOS build compatibility. Place each 
dependency on the verification build machine in a separate directory such as

/usr/local/gdal-2.2.3/
/usr/local/geos-3.6.1/

This would make commits fail upfront on Linux if CMake INCLUDES needed 
by macOS were missing.


Tom
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Build error on MacOS with 9e704ebceb06835e2c28db91e17b45ad296bd207

[Tom Elwertowski]

I got this too. See pull request #6015 for details.

Tom


On 1/8/18 5:03 PM, Tim Sutton wrote:

Hi Does anyone else get this? Was building fine yesterday….


[8/12 20.1/sec] Linking CXX shared module PlugIns/qgis/libgrassplugin7.so
FAILED: PlugIns/qgis/libgrassplugin7.so
: && /usr/local/opt/ccache/libexec/clang++ -Wall -Wextra -Wno-long-long 
-Wformat-security -Wno-strict-aliasing -Wno-return-type-c-linkage 
-Wno-overloaded-virtual -Qunused-arguments -fvisibility=hidden 
-fvisibility-inlines-hidden -O2 -g -DNDEBUG -bundle 
-Wl,-headerpad_max_install_names -Qunused-arguments 
-F/Library/Frameworks -o PlugIns/qgis/libgrassplugin7.so 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassaddfeature.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrasseditrenderer.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassplugin.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassselect.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrasstools.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassmapcalc.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassmodule.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassmoduleinput.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassmoduleoptions.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassmoduleparam.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassnewmapset.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassregion.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassutils.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qgsgrassshell.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/BlockArray.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/ColorScheme.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Emulation.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Filter.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/History.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/HistorySearch.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/KeyboardTranslator.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Pty.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/SearchBar.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Screen.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/ScreenWindow.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Session.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/ShellCommand.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/TerminalCharacterDecoder.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/TerminalDisplay.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/Vt102Emulation.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/kprocess.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/konsole_wcwidth.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/kpty.cpp.o src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/kptydevice.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/kptyprocess.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/qtermwidget.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/tools.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassaddfeature.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrasseditrenderer.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassplugin.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassselect.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrasstools.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassmapcalc.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassmodule.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassmoduleinput.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassmoduleoptions.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassmoduleparam.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassnewmapset.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassregion.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassutils.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/moc_qgsgrassshell.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/moc_Emulation.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/moc_Filter.cpp.o 
src/plugins/grass/7/CMakeFiles/grassplugin7.dir/__/qtermwidget/moc_HistorySearch.cpp.o 

Re: [QGIS-Developer] Qt & Python version for QGIS 3

[Tom Elwertowski]

For QtWebKit, I created and used a build subdirectory inside the source 
directory:


export QT5DIR=/Applications/Qt/5.9/clang_64
alias qmake=$QT5DIR/bin/qmake
qmake ../WebKit.pro

For PyQt, in the source directory:

python3 configure.py --qmake=/Applications/Qt/5.9/clang_64/bin/qmake

Both followed by 'make' and 'sudo make install'

Tom

On 6/14/17 8:41 PM, William Kyngesburye wrote:

How did you build QtWebkit 5.9, or PyQt?  I have the Qt 5.9 binaries, built 
QtWebkit 5.9 from the Qt project version of the source, but qmake doesn't see 
it when building PyQt 5.9 (dev).  qmake says:

Project ERROR: Unknown module(s) in QT: webkit
Project ERROR: Unknown module(s) in QT: webkitwidgets

and skips building the webkit parts of PyQt.

The PyQt list wasn't any help.  Other than suggesting trying the annulen source 
for QtWebkit, which I haven't yet.  Apparently the Qt version doesn't get 
webkit security updates, the annulen version does.


On Jun 6, 2017, at 11:07 AM, Tom Elwertowski <telwertow...@comcast.net> wrote:

Hi William,

I rebuilt QGIS using Qt 5.9 and works the same as with Qt 5.8. I also rebuilt 
the Qt dependencies: QtWebKit 5.9.0, QScintilla 2.10, PyQt 5.8.2, QCA 5f18ebc, 
QtKeyChain 0.8.0 and Qwt 6.1.3.

I did not update the QGIS path fixing installation scripts. Doing 'export 
DYLD_LIBRARY_PATH=/Applications/Qt/5.9/clang_64/lib' and launching from the 
command line gets around this.

qt.conf must contain:
[Paths]
Prefix=/Applications/Qt/5.9/clang_64

Tom
LinkedIn: https://www.linkedin.com/in/thomas-elwertowski-a0886032


On 6/5/17 4:46 PM, William Kyngesburye wrote:

I'm starting to look at building QGIS 3, and I'd like to know what recent 
versions of Qt5 and Python 3 are supported.  I see Qt is up to 5.9 and Python 
is up to 3.6.
-
William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
http://www.kyngchaos.com/
Theory of the Universe
There is a theory which states that if ever anyone discovers exactly what the 
universe is for and why it is here, it will instantly disappear and be replaced 
by something even more bizarrely inexplicable.  There is another theory which 
states that this has already happened.
-Hitchhiker's Guide to the Galaxy 2nd season intro
___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


-
William Kyngesburye <kyngchaos*at*kyngchaos*dot*com>
http://www.kyngchaos.com/

"Time is an illusion - lunchtime doubly so."

- Ford Prefect




___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Qt & Python version for QGIS 3

[Tom Elwertowski]

Hi William,

I rebuilt QGIS using Qt 5.9 and works the same as with Qt 5.8. I also 
rebuilt the Qt dependencies: QtWebKit 5.9.0, QScintilla 2.10, PyQt 
5.8.2, QCA 5f18ebc, QtKeyChain 0.8.0 and Qwt 6.1.3.


I did not update the QGIS path fixing installation scripts. Doing 
'export DYLD_LIBRARY_PATH=/Applications/Qt/5.9/clang_64/lib' and 
launching from the command line gets around this.


qt.conf must contain:
[Paths]
Prefix=/Applications/Qt/5.9/clang_64

Tom
LinkedIn: https://www.linkedin.com/in/thomas-elwertowski-a0886032


On 6/5/17 4:46 PM, William Kyngesburye wrote:

I'm starting to look at building QGIS 3, and I'd like to know what recent 
versions of Qt5 and Python 3 are supported.  I see Qt is up to 5.9 and Python 
is up to 3.6.

-
William Kyngesburye 
http://www.kyngchaos.com/

Theory of the Universe

There is a theory which states that if ever anyone discovers exactly what the 
universe is for and why it is here, it will instantly disappear and be replaced 
by something even more bizarrely inexplicable.  There is another theory which 
states that this has already happened.

-Hitchhiker's Guide to the Galaxy 2nd season intro


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [QGIS-Developer] Qt & Python version for QGIS 3

[Tom Elwertowski]

I have built current QGIS using Qt 5.8 on macOS 10.12 with your 
libraries if available and commandline make for others. I use Python 3.6 
and commandline pip3 for required modules. Qt 5.9 is just out but I will 
give it a try.


Tom
LinkedIn: https://www.linkedin.com/in/thomas-elwertowski-a0886032

On 6/5/17 4:46 PM, William Kyngesburye wrote:

I'm starting to look at building QGIS 3, and I'd like to know what recent 
versions of Qt5 and Python 3 are supported.  I see Qt is up to 5.9 and Python 
is up to 3.6.

-
William Kyngesburye 
http://www.kyngchaos.com/

Theory of the Universe

There is a theory which states that if ever anyone discovers exactly what the 
universe is for and why it is here, it will instantly disappear and be replaced 
by something even more bizarrely inexplicable.  There is another theory which 
states that this has already happened.

-Hitchhiker's Guide to the Galaxy 2nd season intro


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


___
QGIS-Developer mailing list
QGIS-Developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] Failed to build QGIS master on OSX

[Tom Elwertowski]

Hi Matthias, Larry, et al.,

Restoring the iostream header removed from qgspoint.h in 520e3c0 made my 
Mac happy.


Tom Elwertowski


On 2/28/17 3:17 PM, Larry Shaffer wrote:

Hi Tom, et al.,

On Tue, Feb 28, 2017 at 11:41 AM, Tom Elwertowski
<telwertow...@comcast.net <mailto:telwertow...@comcast.net>> wrote:

Hi Larry, Ismail,

I don't use Homebrew but still have this problem.

QGIS c4bc154 built and ran for me on macOS 10.12.3, Qt 5.8, SIP
4.19.1 PyQt 5.8, QScintilla 2.10. The problem seems to arise from a
change made to QgsPoint a few days ago.

My guess is that for macOS, QgsPoint needs an extra or different C++
library header or else a dependency on a newer compiler version has
been introduced.


Matthias, it looks like you worked on that class most recently in
520e3c0 and 1182816. Any ideas?

Regards,

Larry Shaffer
Dakota Cartography
Black Hills, South Dakota
--
Boundless Desktop and QGIS Support/Development
Boundless Spatial - http://boundlessgeo.com <http://boundlessgeo.com/>
lshaf...@boundlessgeo.com <mailto:lshaf...@boundlessgeo.com>


Tom Elwertowski


On 2/28/17 12:26 PM, Larry Shaffer wrote:

Hi Ismail,

Due to this issue:
https://github.com/qgis/homebrew-qgisdev/issues/22
<https://github.com/qgis/homebrew-qgisdev/issues/22>

You may have compilation and runtime problems, which *may* be
unrelated
to QGIS source code. Until the associated Qt-based formulae are
fixed
upstream at Homebrew, this will persist.

Regards,

Larry Shaffer
Dakota Cartography
Black Hills, South Dakota
--
Boundless Desktop and QGIS Support/Development
Boundless Spatial - http://boundlessgeo.com
lshaf...@boundlessgeo.com <mailto:lshaf...@boundlessgeo.com>
<mailto:lshaf...@boundlessgeo.com
<mailto:lshaf...@boundlessgeo.com>>

On Tue, Feb 28, 2017 at 5:00 AM, Ismail Sunni
<imajimat...@gmail.com <mailto:imajimat...@gmail.com>
<mailto:imajimat...@gmail.com <mailto:imajimat...@gmail.com>>>
wrote:

Hi everyone

Today I tried to build QGIS on OSX by following the tutorial
in this
README


(https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md

<https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md>


<https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md

<https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md>>).
I pull the latest commit from QGIS also (sha: f0489c9fc6).
But, I
got this error:

/Users/ismailsunni/dev/cpp/QGIS/src/core/qgspoint.h:287:
error:
invalid operands to binary expression ('std::ostream' (aka
'basic_ostream') and 'char *')
  os << p.toString().toLocal8Bit().data();
  ~~ ^  ~


and

/Users/ismailsunni/dev/cpp/QGIS/src/core/qgsrectangle.h:228:
error: invalid operands to binary expression ('std::ostream'
(aka 'basic_ostream') and 'char *')
  return os << r.toString().toLocal8Bit().data();
 ~~ ^  ~


Does anyone have experience about this also? I have asked
Tim and
Etienne, they also got the same problem.

Best Regards.

--
Ismail Sunni
ismailsunni.wordpress.com <http://ismailsunni.wordpress.com>
<http://ismailsunni.wordpress.com
<http://ismailsunni.wordpress.com>>

___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
<mailto:Qgis-developer@lists.osgeo.org>
<mailto:Qgis-developer@lists.osgeo.org
<mailto:Qgis-developer@lists.osgeo.org>>
List info:
https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>
<https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>>
Unsubscribe:
https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>
<https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>>




___
Qgis-developer mailing list
Qg

Re: [Qgis-developer] Failed to build QGIS master on OSX

[Tom Elwertowski]

Hi Larry, Ismail,

I don't use Homebrew but still have this problem.

QGIS c4bc154 built and ran for me on macOS 10.12.3, Qt 5.8, SIP 4.19.1 
PyQt 5.8, QScintilla 2.10. The problem seems to arise from a change made 
to QgsPoint a few days ago.


My guess is that for macOS, QgsPoint needs an extra or different C++ 
library header or else a dependency on a newer compiler version has been 
introduced.


Tom Elwertowski


On 2/28/17 12:26 PM, Larry Shaffer wrote:

Hi Ismail,

Due to this issue:
https://github.com/qgis/homebrew-qgisdev/issues/22

You may have compilation and runtime problems, which *may* be unrelated
to QGIS source code. Until the associated Qt-based formulae are fixed
upstream at Homebrew, this will persist.

Regards,

Larry Shaffer
Dakota Cartography
Black Hills, South Dakota
--
Boundless Desktop and QGIS Support/Development
Boundless Spatial - http://boundlessgeo.com
lshaf...@boundlessgeo.com <mailto:lshaf...@boundlessgeo.com>

On Tue, Feb 28, 2017 at 5:00 AM, Ismail Sunni <imajimat...@gmail.com
<mailto:imajimat...@gmail.com>> wrote:

Hi everyone

Today I tried to build QGIS on OSX by following the tutorial in this
README
(https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md

<https://github.com/qgis/homebrew-qgisdev/blob/master/development/README.md>).
I pull the latest commit from QGIS also (sha: f0489c9fc6). But, I
got this error:

/Users/ismailsunni/dev/cpp/QGIS/src/core/qgspoint.h:287: error:
invalid operands to binary expression ('std::ostream' (aka
'basic_ostream') and 'char *')
  os << p.toString().toLocal8Bit().data();
  ~~ ^  ~


and

/Users/ismailsunni/dev/cpp/QGIS/src/core/qgsrectangle.h:228:
error: invalid operands to binary expression ('std::ostream'
(aka 'basic_ostream') and 'char *')
  return os << r.toString().toLocal8Bit().data();
 ~~ ^  ~


Does anyone have experience about this also? I have asked Tim and
Etienne, they also got the same problem.

Best Regards.

--
Ismail Sunni
ismailsunni.wordpress.com <http://ismailsunni.wordpress.com>

___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org <mailto:Qgis-developer@lists.osgeo.org>
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
<https://lists.osgeo.org/mailman/listinfo/qgis-developer>




___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer


___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer
Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] CMake cleanup and OS X frameworks

[Tom Elwertowski]

I think we're almost there. Two info.plist.in files still need to be 
added to the top mac directory.


Tom


William Kyngesburye wrote:

D'oh! Sorry about that.  shoulda cherry-picked.  Try it now.

On Jul 5, 2011, at 6:52 PM, John C. Tull wrote:


I had this issue also, worked around it by copying that directory over. I don't 
know if this is related, but trying to access the preferences now brings up the 
shortcut dialog. Settings --  Options gets to preferences, though.

Personally, I'd love to see the menus get cleaned up so that qgis 
looks/feels/acts more like a mac application on mac platforms. This is a 
problem that arose sometime after 1.4 or 1.5, but it is now becoming more than 
a distraction. Multiple, redundant (you like that irony?) menu items, 
disorganized menu structures, Help menu not last in the menu bar, etc.

I wish I knew how to poke around in the ui files to try and fix this, but I do 
not. So no solutions provided here, only complaints. Sorry about that.

Cheers,
John

On Jul 5, 2011, at 3:50 PM, Tom Elwertowski wrote:


Hi William,

I encountered the following error using your update:

CMake Error at images/icons/CMakeLists.txt:12 (ADD_SUBDIRECTORY):
add_subdirectory given source mac which is not an existing directory.

It looks like icon files weren't moved from src/mac to images/icons/mac.

Tom


-
William Kyngesburyekyngchaos*at*kyngchaos*dot*com
http://www.kyngchaos.com/

The equator is so long, it could encircle the earth completely once.




___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] CMake cleanup and OS X frameworks

[Tom Elwertowski]

Hi William,

I encountered the following error using your update:

CMake Error at images/icons/CMakeLists.txt:12 (ADD_SUBDIRECTORY):
  add_subdirectory given source mac which is not an existing directory.

It looks like icon files weren't moved from src/mac to images/icons/mac.

Tom


William Kyngesburye wrote:

I did some CMake cleanup, along with changing the qgis libraries to frameworks 
for OS X.  Some notes:

- removed a lot of unnecessary conditionals for target configuration for 
platform-specific options.  Cmake is designed so that platform-specific options 
are only used for the platform, so they can all be specified in a single 
configuration command. ie for an application:

   ADD_EXECUTABLE ([exename] MACOSX_BUNDLE WIN32 [sources...])

- header installation is an automatic part of targets.  Though some headers may 
still need a separate install command (ui headers don't exist when cmake runs). 
 I also put all global header installation (qgsconfig.h, qgisplugin.h, 
qgsrendererplugin.h) in qgis_core, since in the OS X framework setup there is 
no place for random headers.

- core, gui, analysis, grass and sqlanywhere libs are now frameworks on OS X.  
I couldn't make qgispython a framework because it's loaded dynamically and the 
Qt dyld functions expect standard file extensions, and a framework has none 
(the binary inside the framework, that is).  Really, grass and sqlanywhere 
don't need to be frameworks, maybe I got carried away there ;)

There is also an option to install core, gui and analysis as developer 
frameworks.  This gets around the potential problem of the internal relative path 
linking, and makes the libraries more accessible.  The cmake option is 
QGIS_MACAPP_INSTALL_DEV (bool).  They are installed by default in /Library/Frameworks, 
but this can be changed with QGIS_MACAPP_DEV_PREFIX.

Note that the headers are not updated to make proper frameworks, so -I flags to 
each framework's Headers folder are still needed.  (A proper framework header references other 
framework headers prefixed by the framework name, ie #includeqgis_core/qgis.h.  Needs 
some discussion about reorganizing the headers in QGIS, if there is interest.)

-
William Kyngesburyekyngchaos*at*kyngchaos*dot*com
http://www.kyngchaos.com/

The beast is actively interested only in now, and, as it is always now and always 
shall be, there is an eternity of time for the accomplishment of objects.

- the wisdom of Tarzan





___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer



___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] cmake cleanup to test (all platforms)

[Tom Elwertowski]

The Python library is loading again. Your revisions work for my OS X build.

Tom

William Kyngesburye wrote:

Indeed, I had to change qgispython back to a library - the runtime loading uses 
QLibrary, which looks for a dylib and would need an explicit path to find a 
framework, too much hassle.

___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] 1 Mac section was skipped in r15383

[Tom Elwertowski]

Removing the windowTitle property from qgisapp.ui works for me. The 
title after startup is Quantum GIS r15387M so it's still being set 
somewhere later.


Tom


Martin Dobias wrote:

On Tue, Mar 8, 2011 at 5:29 PM, Tom Elwertowski
telwertow...@comcast.net  wrote:

The following workaround may shed some light on the source of the problem:

Comment out the first line of retranslateUI in ui_ggisapp.h
(MainWindow-setWindowTitle)

This suggests to me that something needed by setWindowTitle for this object
may not be initialized yet.


Hmm... would it help to completely remove windowTitle property from
main window in qgisapp.ui file?

Martin



___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer

Re: [Qgis-developer] svn_marker?

[Tom Elwertowski]

Hi William,

I haven't done a release build since 1.0 but my recollection is that it 
had something to do with putting a useful version string into the Mac 
bundle when building from a tarball without an svn connection.


I believe the svn versioning code has been modified since I last did a 
release build; perhaps a Mac tarball build was not tested at the time. 
If you can successfully build from a tarball (or no longer support 
them), you probably don't need to worry about this.


Tom

William Kyngesburye wrote:

When I overhauled the Mac cmake build, there was one part I thought was odd but 
I left it alone.  The src/mac/contents info.plist custom target has a 
SVN_MARKER as the dependency.  This works in SVN source, but now in the release 
source it doesn't work.  There is an error that there is no rule to make 
SVN_MARKER-NOTFOUND.

What does it do?  I found it used in some other projects, where it seems to be 
some dummy dependency

A search turned up some SVN_MARKER usage in old pre-1.0 versions of the source, 
so I wonder if it was something where a better way to do what it does was found 
but this Mac target was skipped?

It could be a moot point later, since I've found that there is a better way to 
package the Mac application bundle built into cmake now, but the current method 
must be fixed for release until I can figure that out.

-
William Kyngesburyekyngchaos*at*kyngchaos*dot*com
http://www.kyngchaos.com/

Time is an illusion - lunchtime doubly so.

- Ford Prefect


___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer



___
Qgis-developer mailing list
Qgis-developer@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/qgis-developer