Re: Can't build egcs-20000417 with cygwin-1.1.0

["Tim Prince" <tprince at computer dot org>]

I have been building the snapshots by copying the definition of INT_ASM_OP from the interix.h into cygwin.h.  I have no idea how to complete the introduction of this macro for all semi-supported targets.  Someone left this partly done and never finished the job.

 

You must also correct libf2c/libU77/[de]time_.c for consistency with the current Windows API emulations.

 

I have not been able to solve all the problems with the building of .info files.  There are \r characters in one of the original .texi files, and patch introduces \r when a .texi file is patched.  Even when I have been able to fix it up so the .info files get built automatically, they don't install correctly.  The short work-around, of course, is 'touch gcc/gcc.info gcc/f/g77.info'

 

libobjc currently is broken, gcov has always been broken for cygwin, chill is slightly broken (probably trivial).

 

In order to run testsuite, I remove dbz from the list of test directories.  The tests there are designed to fail under cygwin, and have no bearing on the test_summary reports.

 

I suppose Mumit Khan is the only person with enough credibility to get fixes accepted.  I think his primary interest is to support more Windows-specific capabilities than the gcc maintainers are willing to accept.  Still, it would be great if we could somehow get these basic patches in so that we can build a standard compiler suite.

 

I've been unemployed for 7 weeks now and can't count on a good solution, let alone one which would give me time to contribute properly to cygwin or gcc.

----- Original Message -----

From: Christian Jönsson FOA 72

To: gcc@gcc.gnu.org

Sent: Thursday, April 20, 2000 6:40 AM

Subject: Can't build egcs-20000417 with cygwin-1.1.0

I wonder if there might be a chance of being able to build egcs using cygwin-1.1.0

in the near future? I know that it has not been possible during the egcs evolution

before. But, perhaps the time has come to incorporate patches to the egcs

development branch to make it build under cygwin-1.1.0?

 

Right now, I can't get by the following error while bootstrapping the compiler:

 

gcc -c  -DIN_GCC    -g -W -Wall -Wtraditional  -DHAVE_CONFIG_H    -I. -I../../egcs-20000417/gcc -I../../egcs-20000417/gcc/config -I../../egcs-20000417/gcc/../include ../../egcs-20000417/gcc/config/i386/i386.c
../../egcs-20000417/gcc/config/i386/i386.c: In function `incdec_operand':
../../egcs-20000417/gcc/config/i386/i386.c:1143: warning: integer constant is unsigned in ANSI C, signed with -traditional
../../egcs-20000417/gcc/config/i386/i386.c: In function `i386_dwarf_output_addr_const':
../../egcs-20000417/gcc/config/i386/i386.c:2896: `INT_ASM_OP' undeclared (first use in this function)
../../egcs-20000417/gcc/config/i386/i386.c:2896: (Each undeclared identifier is reported only once
../../egcs-20000417/gcc/config/i386/i386.c:2896: for each function it appears in.)
../../egcs-20000417/gcc/config/i386/i386.c: In function `ix86_expand_int_movcc':
../../egcs-20000417/gcc/config/i386/i386.c:5013: warning: integer constant is unsigned in ANSI C, signed with -traditional
../../egcs-20000417/gcc/config/i386/i386.c: In function `ix86_expand_strlensi_unroll_1':
../../egcs-20000417/gcc/config/i386/i386.c:5896: warning: integer constant is unsigned in ANSI C, signed with -traditional
make[2]: *** [i386.o] Error 1

 

 

Cheers,

 

/ChJ