#46441: bob: build failure, possible boost issue ----------------------------------------+--------------------------- Reporter: devans@… | Owner: andre.anjos@… Type: defect | Status: new Priority: Normal | Milestone: Component: ports | Version: Resolution: | Keywords: Port: qt4-mac qt4-x11 boost bob | ----------------------------------------+--------------------------- Changes (by devans@…): * cc: michaelld@…, sean@… (added) * port: qt4-mac qt4-x11 => qt4-mac qt4-x11 boost bob Comment: Replying to [comment:2 and.damore@…]:
See [https://bugreports.qt.io/browse/QTBUG-22829 this Qt bug] for info and possible workarounds: * wrap the boost header (uglier solution) * wrap inclusion of specific boost header in the library's client * switch to Qt5 since the new moc properly expands macros.
Thanks for narrowing this down. I've been trying to understand the inciting event that broke bob, since `bob @1.2.2_6` built correctly but `bob @1.2.2_7` does not. The only difference is a (unrelated) rev bump to bob due to an ABI change in hdf5. I've confirmed that `bob @1.2.2_7` will build without error if boost is reverted locally to `boost @1.56.0_2+no_single+no_static+python27` (r127942). This, in conjunction with the Qt bug report, supports the hypothesis that the problem is triggered by a specific combination of Qt version, boost version and boost includes used in bob. In this case, combination is qt4-mac @4.8.6, boost @1.57.0 and bob @1.2.2. The most recent comment in the Qt bug supports this hypothesis {{{ Only for info: See also with Qt 4.8.6 + boost 1.57.0 + include <boost/signals2.hpp> }}} The best solution would be to fix moc in Qt 4.8.6 but the upstream bug has been closed so a work around in bob may be called for. Copying possibly interested parties. -- Ticket URL: <https://trac.macports.org/ticket/46441#comment:5> MacPorts <https://www.macports.org/> Ports system for OS X