Le 31/05/2011 15:09, Cyrille Lefevre a écrit :
ping
Le 16/03/2011 01:52, Cyrille Lefevre a écrit :
ping
> Le 11/02/2011 21:07, Glenn Fowler a écrit :
> > can you try a build with one of the locales that failed
> > but first edit
> > src/cmd/INIT/package.sh
> > bin/package
> > and change the line
> > LC_ALL=C
> > to
> > LC_ALL=C
> > export LC_ALL
> > this would be a build workaround
> > I still need to check why there are failures for other locales
>
> done
>
> Hi,
>
> I'm back...
>
> so, I've upgraded to cygwin 1.7.8 for a while and the builds seem to
> work much better than ever :-)
> well, it depends on which environment variables are set...
>
> the major drawback is that if NPROC > 1, all builds fails the same way
> whatever the environment variables.
>
> have tried the following builds :
>
> LANG=en_US.UTF-8 by default, LC_ALL isn't set, NPROC defaults to 4.
>
> 1/ LANG=C.UTF-8 LC_ALL=C.UTF-8 bin/package make
> 2/ LC_ALL=C.UTF-8 ...
> 3/ LC_ALL=en_US.UTF-8 ...
> 4/ export LANG=C and LC_ALL=C in bin/package and src/cmd/INIT/package.sh
> 5/ export LANG=C.UTF-8 and LC_ALL=C.UTF-8 ...
>
> ditto, NPROC=1
>
> 11..55 == 1..5
> 66/ 5+1
> 77/ NPROC=1 bin/package make
>
> results :
>
> 1..5 break on cannot send action to coshell or lost contact with coshell
> (i.e. NPROC > 1)
> maybe this is related on
> 44/55/77 break on regular expression..out of space (i.e. patching the
> package command doesn't suffice bcoz of the calling environment)
> 11...33/66 build almost all except cmd/kshlib/dbm_t, cmd/kshlib/dss and
> cmd/kshlib/open
>
> v2$ uname -a
> CYGWIN_NT-6.0 pcvista 1.7.8(0.236/5/3) 2011-03-01 09:36 i686 Cygwin
>
> still vista
>
> v2$ gcc --version
> gcc (GCC) 3.4.4 (cygming special, gdc 0.12, using dmd 0.125)
>
> still 3.4.4
>
> v2$ bin/ksh -c 'echo ${.sh.version}'
> Version JM 93u 2011-02-08
>
> v2$ bin/ksh -c 'type ls'
> ls is a tracked alias for /usr/bin/ls
>
> v2$ bin/ksh -c ls
> arch bin fun include lib man README share src
>
> yeah... no more permission denied !
>
> v2$ bin/package use
>
> v2$ type ls
> ls is a tracked alias for /opt/ast-2011-02-08/arch/cygwin.i386/bin/ls
>
> v2$ ls
> README arch bin fun include lib man share src
>
> yeah again !
>
> v2$ ldd bin/ksh
> ntdll.dll => /cygdrive/c/Windows/system32/ntdll.dll (0x777f0000)
> kernel32.dll => /cygdrive/c/Windows/system32/kernel32.dll
> (0x76b50000)
> cygwin1.dll => /usr/bin/cygwin1.dll (0x61000000)
> ADVAPI32.DLL => /cygdrive/c/Windows/system32/ADVAPI32.DLL
> (0x76ff0000)
> RPCRT4.dll => /cygdrive/c/Windows/system32/RPCRT4.dll (0x775e0000)
> Secur32.dll => /cygdrive/c/Windows/system32/Secur32.dll
> (0x75e10000)
>
> oops! static build ? in other words, why the following line is used :
>
> cc -O -o ksh pmain.o libshell.a ../../../lib/libdll.a
> ../../../lib/libcoshell.a ../../../lib/libcmd.a ../../../lib/libast.a
> /usr/lib/libiconv.a
>
> instead of something like :
>
> cc -O -o ksh pmain.o libshell.a /opt/ast-2011-02-08/bin/cygdll10.dll
> ../../../lib/libcoshell.a /opt/ast-2011-02-08/bin/cygcmd12.dll
> /opt/ast-2011-02-08/bin/cygast54.dll /bin/cygiconv-2.dll
>
> v2$ ldd ksh
> ntdll.dll => /cygdrive/c/Windows/system32/ntdll.dll (0x777f0000)
> kernel32.dll => /cygdrive/c/Windows/system32/kernel32.dll
> (0x76b50000)
> cygwin1.dll => /usr/bin/cygwin1.dll (0x61000000)
> ADVAPI32.DLL => /cygdrive/c/Windows/system32/ADVAPI32.DLL
> (0x76ff0000)
> RPCRT4.dll => /cygdrive/c/Windows/system32/RPCRT4.dll (0x775e0000)
> cygast54.dll => /opt/ast-2011-02-08/bin/cygast54.dll (0x66e80000)
> cygiconv-2.dll => /usr/bin/cygiconv-2.dll (0x674c0000)
> cyggcc_s-1.dll => /usr/bin/cyggcc_s-1.dll (0x6bf40000)
> cygcmd12.dll => /opt/ast-2011-02-08/bin/cygcmd12.dll (0x69240000)
> cygdll10.dll => /opt/ast-2011-02-08/bin/cygdll10.dll (0x6a2c0000)
> Secur32.dll => /cygdrive/c/Windows/system32/Secur32.dll
> (0x75e10000)
>
> PS : builds fail if the underlying path contains variable like
> component, ex. : /opt/ast-2011-02-08_LANG=C.UTF-8_LC_ALL=C.UTF-8_NPROC=1
> the reason I named them /opt/ast-2011-02-08.1, etc.
Regards,
Cyrille Lefevre
--
mailto:[email protected]
_______________________________________________
ast-users mailing list
[email protected]
https://mailman.research.att.com/mailman/listinfo/ast-users
