Thanks, Sean. The preferred way to submit patches is via pull requests on GitHub (http://klee.github.io/klee/developers-guide.html#github), but I'm happy to incorporate this small patch manually if you prefer.
Cristian On 23/09/2014 04:08, Sean Bartell wrote: > Hello, > > Daniel Liew on 2014-09-17: >> On 17 September 2014 20:10, Mark R. Tuttle <[email protected]> wrote: >>> I just did a "git pull" to grab the latest commit (2497fdc) and ran into >>> trouble building against LLVM 3.3 and running the regression tests. >>> >>> Build failed in file src/tools/klee/main.cpp in function >>> KleeHandler::openOutputFile >>> >>> #if LLVM_VERSION_CODE >= LLVM_VERSION(3,5) >>> f = new llvm::raw_fd_ostream(path.c_str(), Error, llvm::sys::fs::F_None); >>> #elif LLVM_VERSION_CODE >= LLVM_VERSION(3,0) >>> f = new llvm::raw_fd_ostream(path.c_str(), Error, >>> llvm::sys::fs::F_Binary); >>> #else >>> f = new llvm::raw_fd_ostream(path.c_str(), Error, >>> llvm::raw_fd_ostream::F_Binary); >>> #endif >>> >>> because llvm::sys::fs::F_Binary was not defined. LLVM 3.3 source code seems >>> to define llvm::raw_fd_ostream::F_Binary and not llvm::sys::fs:F_Binary. I >>> changed the reference to LLVM version 3.0 to 3.4 and the build succeeded. >> >> We are currently targeting LLVM2.9 and LLVM3.4 [1] and are not testing >> LLVM3.3 build support. Do you need to use LLVM3.3 for any particular >> reason? >> >> We'd happily accept a patch to fix your compilation error provided it >> doesn't break the configurations we're testing right now. > > I had the same problem a while ago, and it turns out the correct guard > is >= LLVM_VERSION(3,4). I've attached a patch. > > Thanks, > Sean Bartell > > > > _______________________________________________ > klee-dev mailing list > [email protected] > https://mailman.ic.ac.uk/mailman/listinfo/klee-dev > _______________________________________________ klee-dev mailing list [email protected] https://mailman.ic.ac.uk/mailman/listinfo/klee-dev
