Compiling qt reports only unsupported

[TheOneRing]

CCLASH_TRACKER_MODE=yes And configure called with -no-pch

Using cmake with a Qt based project also results only in unsupported.

Would be great to be able to use cclash with Qt.

[inorton]

can you give me an example of one whole command line?

[TheOneRing]

[36/203] R:\bin\cl.exe /nologo /TP -DLIBSNORE_PLUGIN_PATH=\"r:/plugins/libsnore-qt5\" -DQT_CORE_LIB -DQT_GUI_LIB -DQT_NETWORK_LIB -DQT_NO_CAST_FROM_ASCII -DQT_NO_CAST_FROM_BYTEARRAY -DQT_NO_CAST_TO_ASCII -DQT_NO_SIGNALS_SLOTS_KEYWORDS -DQT_NO_URL_CAST_FROM_STRING -DQT_STRICT_ITERATORS -DQT_USE_FAST_OPERATOR_PLUS -DQT_USE_QSTRINGBUILDER -DSNORE_SUFFIX=\"-qt5\" -DUNICODE -DWIN32_LEAN_AND_MEAN -DWINVER=0x0600 -D_CRT_NONSTDC_NO_DEPRECATE -D_CRT_SECURE_NO_DEPRECATE -D_CRT_SECURE_NO_WARNINGS -D_SCL_SECURE_NO_WARNINGS -D_UNICODE -D_USE_MATH_DEFINES -D_WIN32_IE=0x0600 -D_WIN32_WINNT=0x0600 -Dlibsnore_EXPORTS -Isrc\libsnore -IQ:\snorenotify\src\libsnore -IQ:\snorenotify\src -Isrc -IR:\include\qt5 -IR:\include\qt5\QtCore -IR:\.\mkspecs\win32-msvc2015 -IR:\include\qt5\QtGui -IR:\include\qt5\QtNetwork /DWIN32 /D_WINDOWS /W3 /GR /EHsc /wd4250 /wd4251 /wd4396 /wd4661 /D_DEBUG /MDd /Zi /Ob0 /Od /RTC1 /showIncludes /Fosrc\libsnore\CMakeFiles\libsnore.dir\plugins\plugincontainer.cpp.obj /Fdsrc\libsnore\CMakeFiles\libsnore.dir\ /FS -c Q:\snorenotify\src\libsnore\plugins\plugincontainer.cpp

But I guess the easiest way to test it is to download qt and to try to build an example application

[inorton]

Many thanks. There are quite a few flags there I don't recognise

[inorton]

Can you try setting CCLASH_Z7_OBJ=yes in your environment? cclash can't currently cache builds that use PDB files (especially those with implicit names)

[TheOneRing]

Ouch I confused the completely unrelated CCLASH_TRACKER_MODE=yes with CCLASH_Z7_OBJ=yes . It really looks like it works with cmake and qt. Bootstrapping of qtbase won't work out of the box as it expects a pdb and tries to delete it.

Using ninja (multijob build tool) with cmake killed the cclash server.

[inorton]

Aha I shal investigate!

You might have more luck with tracker mode turned off. It's quite a new feature and is actually slower than normal direct mode. On 28 Jun 2016 14:32, "Hannah von Reth" notifications@github.com wrote:

Ouch I confused the completely unrelated CCLASH_TRACKER_MODE=yes with CCLASH_Z7_OBJ=yes . It really looks like it works with cmake and qt. Bootstrapping of qtbase won't work out of the box as it expects a pdb and tries to delete it.

Using ninja (multijob build tool) with cmake killed the cclash server.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/inorton/cclash/issues/27#issuecomment-229049605, or mute the thread https://github.com/notifications/unsubscribe/AADmGnq-4p-O_kv827AOj0q-UaI1bp3sks5qQSKHgaJpZM4I8hKZ .

[inorton]

On further investigation caching with PDBs is quite problematic, infact I'm suprised anything other than a concurrent msbuild being able to cope with it as mspdbsrv will create or update PDB files often considerably later after the compiler has exited and built a file. This can be very hard to monitor. Eg

• compile file with pdb
• compiler exits, object created, pdb missing
• pdb appears

This can get very wierd if the monent the compiler exits we compile another file that will contribute to the same pdb. If the pdb exists then there are two possible states it can be in before compiling the second file depending on how long we wait..