This patch looks good to me. If you have commit privileges please go ahead and commit it. Otherwise I'll do it for you.
> On Jan 25, 2015, at 7:40 AM, Thomas Hauth <[email protected]> wrote: > > Dear clang Developers, > this is a follow-up on a bug report [3], as I encountered the same > problem. The ccc-analyzer script used during the scan-build static > analysis process is not properly forwarding options to the underlying > compiler. The '-isystem' flag, which marks an include directory to > contain system headers, is only forwarded if a white space is present > between "-isystem" and <path>. While both clang rev 226467 and gcc 4.8.3 > support both ("-isystem <path>" and "-isystem<path>"), the ccc-analyzer > can only handle version with a white space. > > There is no clear documentation whether "-isystem" can contain a > whitespace, but the gcc documentation states, that "-isystem" can be > used in place of "-I" [1] which suggest to use it without a whitespace. > Clang's user manual has one example on using "-isystem", but with a > white space [2]. I added an example makefile with instructions how to > reproduce this problem to the above bug report. > > I attached a patch which enables ccc-analyzer to accept and forward both > the "-isystem <path>" and "-isystem<path>" variant. > > Thanks, > Thomas > > The attached patch was created against 227046. > > Affected Files: > tools/scan-build/ccc-analyzer > > [1] > https://gcc.gnu.org/onlinedocs/gcc/Directory-Options.html > [2] > http://clang.llvm.org/docs/UsersManual.html#controlling-diagnostics-in-system-headers > [3] > http://llvm.org/bugs/show_bug.cgi?id=13237 > <ccc-analyzer-isystem-no-whitespace.patch>_______________________________________________ > cfe-commits mailing list > [email protected] > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
