Version: (using KDE KDE 3.1.3) Installed from: Compiled From Sources Compiler: gcc-3.3..2 OS: Linux In doctreeview/doctreeviewwidget.cpp, line 766, QString is implicitly cast to (const char*): if ( (f = fopen(d.filePath(*it), "r")) != 0) However, there's no appropriate casting operator available (I think that depends on the version of Qt or something; the fact remains that there are configurations without such an operator).
This in not a bug! This is a compilation error. Please: -Report bugs to the bug database (here) -Report compilation erros to the mailing list (not here) PS: How about providing you QT, autoconf and automake versions?
> There's a part of the code that tries to access an (at least in some > configurations) non-existing function, and you call that not a bug? > Considering that in another programming language, this might actually be a > runtime error, it's kind of a stretch to call it "not a bug". I use KDE 3.1.3 , QT 3.1.2 and autoconf 2.57 just like you do. BUT I have automake 1.7.5 and gcc 2.95.3. There are developers that have tested gcc 3.2.x and 3.3.x in all their flavors and variations so gcc version is not an issue here. But your outdated automake version is. Have you used the trick that I recomended you in the other mail to bypass the need for an uptodate automake? If not, please do it and if the problem presists please update automake to 1.6.1 at least. > Well duh. I really needed that I code for you for free and you still complain about it! I really, really needed that!
Subject: Re: kdevelop 3.0.0a6 fails to compile (doctreeview/doctreeviewwidget.cpp) On Saturday 13 September 2003 14:56, you wrote: > ------- You are receiving this mail because: ------- > You reported the bug, or are watching the reporter. > > http://bugs.kde.org/show_bug.cgi?id=64103 > > > > > ------- Additional Comments From a.lucas@tu-bs.de 2003-09-13 14:56 > ------- > > > There's a part of the code that tries to access an (at least in some > > configurations) non-existing function, and you call that not a bug? > > Considering that in another programming language, this might actually > > be a runtime error, it's kind of a stretch to call it "not a bug". > > I use KDE 3.1.3 , QT 3.1.2 and autoconf 2.57 just like you do. > BUT I have automake 1.7.5 and gcc 2.95.3. > There are developers that have tested gcc 3.2.x and 3.3.x in all their > flavors and variations so gcc version is not an issue here. But your > outdated automake version is. Have you used the trick that I recomended > you in the other mail to bypass the need for an uptodate automake? If > not, please do it and if the problem presists please update automake to > 1.6.1 at least. Again, this doesn't seem to have anything to do with the automake version. I checked out the latest CVS anyway, did a make -f Makefile.cvs (with automake updated to 1.6.3), and I get exactly the same results. What it boils down to is this: configure sets -DQT_NO_ASCII_CAST, which masks the casting operator in qstring.h I tracked QT_NO_ASCII_CAST down to admin/acinclude.m4.in, where it is set in the default KDE CXXFLAGS. There doesn't seem to be any part of the configure script that might undefine QT_NO_ASCII_CAST. It's weird. Looks like there is no way QT_NO_ASCII_CAST could be unset in the configure script. Sure, compilation works on your end, but if - -DQT_NO_ASCII_CAST is set for you as well, the Qt headers must be different somehow. I can't think of a non-weird explanation for this. However, grep QT_NO_ASCII_CAST ChangeLog suggests that previous changes have been made to support this condition. The rationale is probably that a simple cast cannot express which character set you want (plain old ASCII, UTF-8, whatever). [snip potential flamewar ;) - my apologies if my earlier reply sounded too harsh] cu, Nicolai -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.3 (GNU/Linux) iD8DBQE/YzMpsxPozBga0lwRAl+BAKDO0gPhE3+Jr3X7bCTVmOwmA1lOVACePLWm qbS4bZJwlvsfUSvojlvpLVU= =BObT -----END PGP SIGNATURE-----