On Oct 25, 2012, at 11:40 AM, Joerg Sonnenberger <[email protected]> wrote:
> Hi all, > this is the second version of the patch. A small part already went into > the tree (PrintFatalError and the clang changes, except > CMakeLists.txt/Makefile). > > The one really nasty part of this change is CodeGenDAGPattern.cpp. I've > changed the TreePattern code to provide a sticky marker. The first error > hit during the resolution process is printed and it tries to bail out > from that. I am not completely sure if the current logic is aggressive > enough or whether hasError() checks are needed in a few more cases. > Since this only matters for at least somewhat broken specifications, it > shouldn't be a problem to add more checks as needed. LGTM. /jakob _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
