On Thu, Jul 21, 2016 at 7:23 PM, David Malcolm <dmalc...@redhat.com> wrote: > On Thu, 2016-07-21 at 10:38 -0400, David Malcolm wrote: >> On Thu, 2016-07-21 at 10:28 -0400, David Malcolm wrote: >> > On Thu, 2016-07-21 at 09:00 +0200, Uros Bizjak wrote: >> > > Hello! >> > > >> > > > > gcc/ChangeLog: >> > > > > PR c/70339 >> > > > > * diagnostic-show-locus.c (diagnostic_show_locus): If this is >> > > > > the >> > > > > same location as last time, don't skip if we have fix-it >> > > > > hints. >> > > > > Clarify the skipping logic by converting it from one "if" >> > > > > clause >> > > > > to repeated "if" clauses. >> > > > > * spellcheck-tree.c: Include "cpplib.h". >> > > > > (find_closest_macro_cpp_cb): Move here from c/c-decl.c. >> > > > > (best_macro_match::best_macro_match): New constructor. >> > > > > * spellcheck-tree.h (struct edit_distance_traits<cpp_hashnode >> > > > > *>): >> > > > > Move here from c/c-decl.c. >> > > > > (class best_macro_match): Move here from c/c-decl.c, >> > > > > converting >> > > > > from a typedef to a subclass, gaining a ctor. >> > > >> > > This part introduced bootstrap failure on CentOS 5.11: >> > > >> > > g++ -std=gnu++98 -fno-PIE -c -g -DIN_GCC -fno-strict >> > > -aliasing >> > > -fno-exceptions -fno-rtti -fasynchronous-unwind-tables -W -Wall >> > > -Wwrite-strings -Wcast-qual -Wno-format -Wmissing-format >> > > -attribute >> > > -Woverloaded-virtual -fno-common -DHAVE_CONFIG_H -I. -I. >> > > -I../../git/gcc/gcc -I../../git/gcc/gcc/. >> > > -I../../git/gcc/gcc/../include >> > > -I../../git/gcc/gcc/../libcpp/include >> > > -I/usr/local/include -I/usr/local/include -I/usr/local/include >> > > -I../../git/gcc/gcc/../libdecnumber >> > > -I../../git/gcc/gcc/../libdecnumber/bid -I../libdecnumber >> > > -I../../git/gcc/gcc/../libbacktrace -o spellcheck-tree.o -MT >> > > spellcheck-tree.o -MMD -MP -MF ./.deps/spellcheck-tree.TPo >> > > ../../git/gcc/gcc/spellcheck-tree.c >> > > ../../git/gcc/gcc/spellcheck-tree.c: In constructor >> > > ‘best_macro_match::best_macro_match(tree_node*, edit_distance_t, >> > > cpp_reader*)’: >> > > ../../git/gcc/gcc/spellcheck-tree.c:94: error: class >> > > ‘best_macro_match’ does not have any field named ‘best_match’ >> > > ../../git/gcc/gcc/spellcheck-tree.c:94: error: no matching >> > > function >> > > for call to ‘best_match<tree_node*, cpp_hashnode*>::best_match()’ >> > > ../../git/gcc/gcc/spellcheck.h:91: note: candidates are: >> > > best_match<GOAL_TYPE, CANDIDATE_TYPE>::best_match(GOAL_TYPE, >> > > edit_distance_t) [with GOAL_TYPE = tree_node*, CANDIDATE_TYPE = >> > > cpp_hashnode*] >> > > ../../git/gcc/gcc/spellcheck.h:81: note: >> > > best_match<tree_node*, cpp_hashnode*>::best_match(const >> > > best_match<tree_node*, cpp_hashnode*>&) >> > > gmake: *** [spellcheck-tree.o] Error 1 >> > > >> > > CentOS 5 has pretty old system compiler [gcc version 4.1.2 >> > > 20080704 >> > > (Red Hat 4.1.2-55)], but the build worked until the above patch >> > > was >> > > committed. Is it possible to work around this failure somehow? >> > >> > Sorry about the breakage. >> > >> > I've now built a gcc 4.1 locally, and with that I can reproduce the >> > issue. Am investigating... >> >> The following patch seems to fix it; successfully compiles that file >> with my local builds of gcc 7 and gcc 4.1 (and with Fedora system gcc >> 4.8.3). >> >> Does it fix the build for you? >> >> Bootstrap®rtesting underway here. >> >> gcc/ChangeLog: >> * spellcheck-tree.c (best_macro_match::best_macro_match): >> Explicitly specify the template arguments when invoking the >> base >> class constructor, to help older C++ compilers. >> >> Sorry again about the breakage. > > This patch survived bootstrap®rtest, and compiles and runs OK with > gcc-4.1, so I've committed it to trunk as r238598 (under the "obvious" > rule). > > Please let me know if there are any configurations for which it still > doesn't work.
Yes, the patch solves bootstrap problem for me. Thanks! Uros.
