> Sent: Wednesday, December 21, 2016 at 1:48 PM > From: "william.croc...@analog.com" <william.croc...@analog.com> > To: interest@qt-project.org > Subject: Re: [Interest] Frustrated, Why doesn't Qt ever compile? > > On 12/21/2016 01:33 PM, Jason H wrote: > > When I compile Qt, it is a harrowing experience. I've _never_ actually had > > a compile go as planned, which "as planned" is: configure && make && make > > install (with configure options of course). I don't think it's me. I > > thought that with all the gerrit review, sanity bots, COIN, that these > > problems would be a thing of the past? I'm trying to understand why/how > > it's so hard to compile Qt? > > > > It was hard for me because I lack some of the necessary libraries or mine > were out of date. So I had to iterate.
The configure script should pick that up though. One enhancement that I'd appreciate is configure recording it's command line somewhere so I know how a build qt was previously configured. > Also it took a while to realize that some builds omit the 'make install' step. > It might have been smarter to convert 'make install' to a noop. > > 1 : configure; make; fail on this. > 2 : configure -nothis; make; fail on that. > 3 : configure -nothis -nothat; make -> fail on other. > 4 : configure -nothis -nothat -noohter; make -> SUCCESS. > 5 : make install; fail in a non obvious way. > 6 : make clean distclean; goto 4 skipping the install step. > 7 : repeat for release build. > 8 : Take a break > 9 : Try to build the doc. > 10 : Give up and use the online doc. That's pretty much my workflow as well. Fortunately, I can usually skip the broken module and still have a build that works for my purposes. > > My most recent frustration is with the freshly minted 5.6.2 source, wherein > > I received: > > > > https://www.mail-archive.com/interest@qt-project.org/msg24196.html > > > >> error: cannot initialize a variable of type 'const CBCentralManagerState' > >> with an rvalue of type 'CBManagerState' > >> const CBCentralManagerState cbState(central.state); > > > > There seems to be a second issue that I will also get as well. Jake did > > supply that issues 1 & 2 were fixed, but: > > > > 1) Fixed by https://codereview.qt-project.org/#/c/172072/. Will be in 5.6.3 > > + > > 5.7.1 + 5.8.0. > > > > Ok, great, I can move to 5.7.1 > > > > 2) Probably https://bugreports.qt.io/browse/QTBUG-55915? The fix is in > > 5.6.2 and > > 5.8.0 but won't be in 5.7 until 5.7.2. > > > > That's not good. There's no 5.7.2 yet. > > > > So I'm kinda stuck as to what to do? > > > > > > In this specific case, this seems to be an iOS10 deprecation, but to get > > back to the main point, it doesn't matter the platform, they all don't > > "just compile". I've got a build of 5.7.0, for raspberry pi that won't > > build, an Android build of 5.6.2 that won't compile as well. Sometime > > issuing multiple make commands seems to fix it, but that's some voodoo > > there. > > > > > > Sometimes, I wonder if it's possible/wish it were possible to get a > > pre-built download of Qt so that for any patches I need to try, I can just > > download that, apply the patch and compile that one object file and it's > > module. It would save me days, and I think it wouldn't take much to capture > > the build from a COIN build. > > > > _______________________________________________ > > Interest mailing list > > Interest@qt-project.org > > http://lists.qt-project.org/mailman/listinfo/interest > > > > > > > _______________________________________________ > Interest mailing list > Interest@qt-project.org > http://lists.qt-project.org/mailman/listinfo/interest > _______________________________________________ Interest mailing list Interest@qt-project.org http://lists.qt-project.org/mailman/listinfo/interest
