I have changed the subject line since the old one was barely relevant any more.
On 2009-06-24 22:58-0700 Alan W. Irwin wrote: > Let's proceed in steps. > > 1. Get everything to work with qt.cpp unsplit both for DYNDRIVERS ON and OFF > and for the new and old build system for the installed examples. I would > also like to see the pyqt4 extension module (which also depends on > libplplotqtd) working for the first time as well. Unless you prefer > something different, tomorrow (Thursday) I will work on this including > checking in your current patch to the unsplit version of qt.cpp (assuming it > works for me as well as it did above for you). > > 2. Split qt.cpp and confirm things still work for the four cases (DYNDRIVERS > ON and OFF and for the new and old build system) and also for the pyqt4 > extension module. I actually think splitting qt.cpp into device driver code > and a library is generally a good idea. However, there are some build > system issues to be aware of for the DYNDRIVERS OFF case, but it should be > fairly straightforward to solve them once we have 1. working in all cases. > > 3. Document. Once, I actually looked at the patch Dmitri sent to the list I had to agree with him that those macros made the source code difficult to understand and essentially emulated splitting the source code. Therefore, I skipped 1. above and proceeded directly to 2 with the aid of a patch that Dmitri sent me off list to split drivers/qt.cpp into drivers/qt.cpp and bindings/qt_gui/plqt.cpp. I added some cmake infrastructure to support the split source code, and the result (revision 10073) gives clean valgrind results for all four kinds of qt devices as well as qt_example. Thanks, Dmitri for your "split" patch, and thank Alban for your earlier patch to fix uninitialized variables. Those two patches have made these completely clean valgrind results possible. The build-tree and install tree versions of qt.so, libplplotqtd, and plplot_pyqt4.so pass the "ldd -r" test for undefined symbols. Nevertheless, the split does not solve the "black/blank gui" problem for pyqt4_test.py. Hopefully, though, the split code is now much more readable and easier to debug for that issue. N.B. The above tests were done just for the default CMake configuration (ENABLE_DYNDRIVERS ON, ENABLE_qt ON) and with the new build system for the installed examples. Additional testing for other configurations is in progress, and in particular there are still a few obvious changes I have to make and test for the traditional build system for the installed examples to complete this work. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------------ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
