Bug #118950 for Syntax-Keyword-Try: Compilation failure on some systems

Thu Nov 24 16:18:26 2016 SREZIC [...] cpan.org - Ticket created

Subject:

Compilation failure on some systems

On some of my smokers I see compilation failures like this: cc -I/usr/lib/arm-linux-gnueabihf/perl/5.20/CORE -DVERSION="0.03" -DXS_VERSION="0.03" -fPIC -c -D_REENTRANT -D_GNU_SOURCE -DDEBIAN -fwrapv -fno-strict-aliasing -pipe -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -O2 -g -o lib/Syntax/Keyword/Try.o lib/Syntax/Keyword/Try.c lib/Syntax/Keyword/Try.xs: In function ârethread_opâ: lib/Syntax/Keyword/Try.xs:143:5: error: âforâ loop initial declarations are only allowed in C99 or C11 mode for(OP *kid = cUNOPx(op)->op_first; kid; kid = OpSIBLING(kid)) ^ lib/Syntax/Keyword/Try.xs:143:5: note: use option -std=c99, -std=gnu99, -std=c11 or -std=gnu11 to compile your code lib/Syntax/Keyword/Try.xs: In function âMY_walk_optree_try_in_evalâ: lib/Syntax/Keyword/Try.xs:185:13: error: redeclaration of âkidâ with no linkage for(OP *kid = cUNOPx(op)->op_first; kid; kid = next) { ^ lib/Syntax/Keyword/Try.xs:184:9: note: previous declaration of âkidâ was here OP *kid, *next, *prev = NULL; ^ lib/Syntax/Keyword/Try.xs:185:5: error: âforâ loop initial declarations are only allowed in C99 or C11 mode for(OP *kid = cUNOPx(op)->op_first; kid; kid = next) { ^ error building lib/Syntax/Keyword/Try.o from 'lib/Syntax/Keyword/Try.c' at /usr/share/perl/5.20/ExtUtils/CBuilder/Base.pm line 175. This happens on linux systems (with Debian/jessie and Debian/wheezy, standard C compiler) and freebsd 9 systems. Compilation is successful on my freebsd 10 system and on Mac OS X (10.11).

Thu Nov 24 17:39:42 2016 leonerd-cpan [...] leonerd.org.uk - Correspondence added

On Thu Nov 24 16:18:26 2016, SREZIC wrote: Show quoted text

> lib/Syntax/Keyword/Try.xs:143:5: error: âforâ loop initial > declarations are only allowed in C99 or C11 mode > for(OP *kid = cUNOPx(op)->op_first; kid; kid = OpSIBLING(kid)) > ^

Bah. It would be really nice if core would fix this. Make sure to pass -std=c89 when compiling with gcc, so it isn't free to pick a default that's different in different places. If we're not going to use C99, then *force* it to be C89 to pick these things up earlier. -- Paul Evans

Thu Nov 24 17:39:42 2016 The RT System itself - Status changed from 'new' to 'open'

Thu Nov 24 19:11:07 2016 leonerd-cpan [...] leonerd.org.uk - Correspondence added

Anyway, patch attached. -- Paul Evans

Thu Nov 24 19:11:29 2016 leonerd-cpan [...] leonerd.org.uk - Correspondence added

Anyway, patch attached. -- Paul Evans

Subject:

rt118950.patch

=== modified file 'lib/Syntax/Keyword/Try.xs' --- lib/Syntax/Keyword/Try.xs 2016-11-24 10:24:24 +0000 +++ lib/Syntax/Keyword/Try.xs 2016-11-25 00:10:28 +0000 @@ -102,7 +102,7 @@ return op; if(warnings == pWARN_STD) - // TODO: understand what STD vs ALL means + /* TODO: understand what STD vs ALL means */ warning_bits = WARN_ALLstring; else if(warnings == pWARN_ALL) warning_bits = WARN_ALLstring; @@ -139,9 +139,11 @@ break; } - if(op->op_flags & OPf_KIDS) - for(OP *kid = cUNOPx(op)->op_first; kid; kid = OpSIBLING(kid)) + if(op->op_flags & OPf_KIDS) { + OP *kid; + for(kid = cUNOPx(op)->op_first; kid; kid = OpSIBLING(kid)) rethread_op(kid, old, new); + } } #define walk_optree_try_in_eval(op_ptr, root) MY_walk_optree_try_in_eval(aTHX_ op_ptr, root) @@ -182,7 +184,7 @@ if(op->op_flags & OPf_KIDS) { OP *kid, *next, *prev = NULL; - for(OP *kid = cUNOPx(op)->op_first; kid; kid = next) { + for(kid = cUNOPx(op)->op_first; kid; kid = next) { OP *newkid = kid; next = OpSIBLING(kid);

Thu Nov 24 19:11:30 2016 leonerd-cpan [...] leonerd.org.uk - Status changed from 'open' to 'patched'

Mon Nov 28 09:34:31 2016 leonerd-cpan [...] leonerd.org.uk - Correspondence added

Released in 0.04 -- Paul Evans

Mon Nov 28 09:34:31 2016 leonerd-cpan [...] leonerd.org.uk - Status changed from 'patched' to 'resolved'

Mon Nov 28 09:34:36 2016 leonerd-cpan [...] leonerd.org.uk - Fixed in 0.04 added