I have eliminated the error: use 'template' keyword to treat 'as' as a dependent template name using this solution
http://stackoverflow.com/a/3786481/66242 g++ compilation went ok no crashes for database creation clang++ compilation went ok but with one crash in help db restoring (I will retest) On Thu, Jun 20, 2013 at 9:58 AM, marius adrian popa <map...@gmail.com> wrote: > I compile with clang++ and indeed there are many warning in 2.5/3.0 > > export CXX="clang++" > ./autogen.sh > > I propose to solve them in 3.0 trunk branch , so we don't introduce > any new bugs in the stable 2.5 branch > > So if you have patches please send them to this list > > with current clang on debian jessie i have this error > > In file included from src/jrd/../jrd/Function.h:28: > src/jrd/../dsql/Nodes.h:508:23: error: use 'template' keyword to treat > 'as' as a dependent template name > return node ? node->as<T>() : NULL; > ^ > template > > > On Thu, Jun 20, 2013 at 9:26 AM, Alex Peshkoff <peshk...@mail.ru> wrote: >> On 06/19/13 20:40, Lee Graber wrote: >>> I am hoping I got the correct discussion alias. I am trying to figure out >>> if I am going down a hole that I will not be happy with. My team uses >>> Firebird's embedded engine for a couple of things and we are currently on >>> the 2.1.x line. We are attempting to build embedded for Mac OSX using >>> clang++ and libc++ but as I have started this I am finding myself fixing a >>> lot of syntactical errors. There are lots of issues from c++11-narrowing >>> checks among others. I can see on some threads that there are developers >>> building using clang but from what I can glean that must only be happening >>> on the 3.0 branch (I tried 2.5.x and it also had issues building so I stuck >>> with 2.1.x to try and fix things since we are not looking to bump versions >>> right now). If 3.0 is building with clang++ and libc++ in 3.0 branch, then >>> someone must have done some work. >> >> As far as I know people used to build at least 2.5 using clang but it >> was more than a year ago. And you know that sometimes warnings in >> compilers tend to become errors in next versions. Therefore 2.5 can be >> not buildable with fresh clang. What about 3.0 - may be it was also >> tried with clang that time, but there were so many changes since that >> time... We use VC on windows and gcc on other systems (there are 2.5 >> ports to native compilers for most of widely used unixes). But what >> about Mac - we always used gcc. >> >>> Does anyone know if this is an achievable goal by just fixing the missing >>> casts, template compilation issues, string concat issues, ... >> >> Definitely yes, but 2.1 is bad choice. That code is about 6 years old >> (when branch was created) and therefore has more suspicious places than >> fresher versions. >> >>> The rest of our product is built using clang and libc++ and we want this >>> embedded component to be built with the same tools. I also have not been >>> able to locate the 3.0 code base (if that is even something that is public >>> domain) >> >> Certainly - license remains the same. >> svn checkout svn://svn.code.sf.net/p/firebird/code/firebird/trunk >> >>> as I was going to compare my changes to those in 3.0 (if the files still >>> exist) just to sanity check my changes. >>> >> >> Such comparison is far not simple. Code was reworked in a lot of places. >> >> Alex. >> >> >> ------------------------------------------------------------------------------ >> This SF.net email is sponsored by Windows: >> >> Build for Windows Store. >> >> http://p.sf.net/sfu/windows-dev2dev >> Firebird-Devel mailing list, web interface at >> https://lists.sourceforge.net/lists/listinfo/firebird-devel ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
