This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Website page states "cvs co -c" ?; gcc in NT croaks using Cygwin winnt.h

I hope you update your CVS pages so that they say how to list
the modules. It seems to be a most obvious thing to document but
most people would find that information very difficult to find.

Please document   'cvs co -c' at the gcc website.
Maybe this is the best page for such a comment:

Or every page including, e.g., the page saying how to get GCC Java via cvs.

Here is a log showing that 'cvs co -c' does list the special options.
That allows people to see all the options before they start and thus
it could improve their opinion of your cvs service, as well as allowing
lower running costs.

                                                                                   10:19:15 in
                                                                                   10:19:22 in
 > cvs login
(Logging in to
CVS password:
cvs login: authorization failed: server rejected access to /cvs/ada for user anonc
                                                                                   10:19:29 in
                                                                                   10:19:52 in
 > export CVSROOT
                                                                                   10:19:56 in
 > cvs login
(Logging in to
CVS password:
                                                                                   10:20:01 in
 > cvs co -c
gcc-ada      -a gcc/gcc/ada
gcc-chill    -a gcc/gcc/ch gcc/libchill
gcc-core     -a !gcc/gcc/testsuite !gcc/gcc/ada !gcc/gcc/ch !gcc/libchill
              !gcc/gcc/cp !gcc/libstdc++ !gcc/libio !gcc/libstdc++-v3 !gcc/gcc/f
              !gcc/libf2c !gcc/gcc/java !gcc/boehm-gc !gcc/libffi !gcc/libjava
              !gcc/zlib !gcc/fastjar !gcc/gcc/objc !gcc/libobjc gcc
gcc-full     -a gcc
gcc-g++      -a gcc/gcc/cp gcc/libstdc++ gcc/libstdc++-v3 gcc/libio
gcc-g++-testsuite -a gcc/gcc/testsuite/g++.old-deja gcc/gcc/testsuite/g++.dg
gcc-g77      -a gcc/gcc/f gcc/libf2c
gcc-g77-testsuite -a gcc/gcc/testsuite/g77.f-torture
gcc-java     -a gcc/gcc/java gcc/boehm-gc gcc/libffi gcc/libjava gcc/zlib
gcc-objc     -a gcc/gcc/objc gcc/libobjc
gcc-testsuite -a !gcc/gcc/testsuite/g++.old-deja !gcc/gcc/testsuite/g++.dg
              !gcc/gcc/testsuite/g77.f-torture gcc/gcc/testsuite
libstdc++-v3 gcc/libstdc++-v3
                                                                                   10:44:08 in

I found that "cvs co -c" was mentioned at this Usenet webpage:

Can the pages be updated to show that "cvs co -c" allows smaller downloads.
That lists "ada" as being part of GCC. The online documentation says all this
which is plainly incomplete and it does not mention Java or FORTRAN and it
does not say that "cvs -z9 co gcc-ada" can entered.

 >CVS tags and checkouts
 >You can check out the latest version of the GCC 3.0 release branch with the following command:
 >cvs -z 9 co -rgcc-3_0-branch gcc
 >By changing the -r argument above you can check out particular releases or snapshots or the latest snapshot. Other tags/branches you may want to check out include:
 >	gcc_latest_snapshot
 >	gcc_ss_yyyymmdd
 >	gcc_3_0_2_release
 >	gcc_3_0_1_release
 >	gcc_3_0_release
 >	gcc-3_0-branch
 >	gcc-2_95_3
 >	gcc-2_95_2-release
 >	gcc-2_95_1-release
 >	gcc-2_95-release
 >	gcc-2_95-branch
 >	egcs_1_1_2_release
 >	egcs_1_1_1_release
 >	egcs_1_1_release
 >	egcs_1_1_branch
 >	egcs_1_0_3_release
 >	egcs_1_0_2_release
 >	egcs_1_0_1_release
 >	egcs_1_0_release
 >	egcs_ss_yyyymmdd For snapshots prior to 19990913
 >	gnu-win32-b20-branch
 >	dfa-branch. This branch contains the insn scheduler with an automaton based
 >         pipeline hazard recognizer.



The cvs website can't be browsed by using the Internet Explorer webbrowser


This is the last console log I got from a compile of GCC. I tried using
some old cygwin header files and that did not fix the problem. Possibly
it got fixed.

The whole compile process depends on files that are not provided by gcc.
The winnt.h file that causes gcc to be unable bootstrap and build itself
is provided by Redhat or Cygwin. I presume the problem is a gcc bug.

../../../../gcc/libstdc++-v3/libsupc++/ warning: -ffunction-sections
    may affect debugging on some targets.
In file included from /usr/include/windef.h:145,
                  from /usr/include/windows.h:98,
                  from /cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3/include/i686-pc-cygwin/bits/gthr-default.h:
                  from /cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3/include/i686-pc-cygwin/bits/gthr.h:98,
                  from ../../../../gcc/libstdc++-v3/libsupc++/
/usr/include/winnt.h:2478: `struct _REPARSE_DATA_BUFFER::<anonymous
    union>::<anonymous>' invalid; an anonymous union can only have non-static
    data members
/usr/include/winnt.h:2485: `struct _REPARSE_DATA_BUFFER::<anonymous
    union>::<anonymous>' invalid; an anonymous union can only have non-static
    data members
/usr/include/winnt.h:2492: `struct _REPARSE_DATA_BUFFER::<anonymous
    union>::<anonymous>' invalid; an anonymous union can only have non-static
    data members
In file included from /usr/include/windows.h:102,
                  from /cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3/include/i686-pc-cygwin/bits/gthr-default.h:
                  from /cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3/include/i686-pc-cygwin/bits/gthr.h:98,
                  from ../../../../gcc/libstdc++-v3/libsupc++/
/usr/include/winbase.h:901: `struct _PROCESS_HEAP_ENTRY::<anonymous
    union>::<anonymous>' invalid; an anonymous union can only have non-static
    data members
/usr/include/winbase.h:905: `struct _PROCESS_HEAP_ENTRY::<anonymous
    union>::<anonymous>' invalid; an anonymous union can only have non-static
    data members
make[4]: *** [eh_alloc.lo] Error 1
make[4]: Leaving directory `/cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3/libsupc++'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3'
make[2]: *** [all-recursive-am] Error 2
make[2]: Leaving directory `/cygdrive/m/gcc-compile/objdir/i686-pc-cygwin/libstdc++-v3'
make[1]: *** [all-target-libstdc++-v3] Error 2
make[1]: Leaving directory `/cygdrive/m/gcc-compile/objdir'
make: *** [bootstrap] Error 2

23:25:56 /cygdrive/m/gcc-compile/objdir

I just re-cvsupped and it looks like the GCC file that was causing the
bootstrap compile of GCC to crash, has not been updated.
I am not really concerned by it but I was wondering if it might compile
in 2002. It looks like it didn't compile in Windows NT in 2001. It seems
to have been taken down by Redhat include files, yet I guess it has been
a problem with gcc code having bugs. The seemingly offending part of
winnt.h was last changed in the approx February 2001 release or so.
So I guess it did not run since then. Gcc also would not compile when
older Cygwin headers were used.

I was using these options.

   $ nice ../gcc/configure --enable-libgcj --enable-threads=posix \
     --prefix=/gccj --enable-shared --enable-languages=c++,java \
     --with-as=/bin/as --with-ld=/bin/ld

   $ nice make bootstrap
   $ nice make
   $ make install

It was dying on the 'make bootstrap' or on the 'make'

I once wrote to a Redhat developer in Brazil who said when gcc does
not compile then it does not compile. It seems to be noncompilable
because of Redhat code and it has been bootstrappable for most of

Hopefully Gcc would move over to Ada which runs well and has
better error messages.

I suppose the person reading this might not be able to do much about
GCC "won't compile - bugs" problems.


The Internet Explorer 5.5SP2 tells the webserver of the GCC CVS website
to gzip the data. Then when MSIE gets the gzipped data, it can't decode
it. Accepting that as my problem, I not that another way to reduce
bandwidth is to document  "cvs co -c" and say that it lists modules.

                                                                                   11:22:57 in
 > ptunnel 85
  Listen = 85,  Remote =
  IP Number of =
  {1+Accept.. {1+Accepted {2-Connect.. {2-Connected (359:
GET HTTP/1.1\r\n
Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/x-gsarcade-launch, */
Accept-Language: en-us\r\n
Accept-Encoding: gzip, deflate\r\n
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 4.0)\r\n
Proxy-Connection: Keep-Alive\r\n
:359) {3+Accept.. [1088:
HTTP/1.0 200 OK\r\n
Connection: close\r\n
Date: Wed, 14 Nov 2001 22:22:12 GMT\r\n
Server: Apache/1.3.19 (Unix)  (Red-Hat/Linux)\r\n
Content-encoding: x-gzip\r\n
Content-Type: text/html\r\n

    [Printed out using the proxy at ]

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]