* build successful (was Re: winsup in EGCS?)
@ 1999-06-20 12:04 Phil Edwards
1999-06-20 14:47 ` Gabriel Galibourg
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Phil Edwards @ 1999-06-20 12:04 UTC (permalink / raw)
To: cygwin
> $ <srcdir>/configure --prefix=/Cygnus/cygwin-b20 \
> --exec-prefix=/Cygnus/cygwin-b20/H-i586-cygwin32 \
> -v i586-cygwin32
>
> $ make bootstrap > make.log 2>&1
I've been trying to use "make blah | tee /something 2&>1" but the error
messages don't get captured in the logfile.
> When it builds, just install it elsewhere:
>
> $ make prefix=/home/EGCS --exec-prefix=/home/EGCS/H-i586-cygwin32 \
> install
Oh. I've never tried to mess around with --exec-prefix. Never seems
to do what I thought it would, and since "uninstall" has never ever
worked, I don't experiment any more (egcs takes nearly six hours to build
on a P133 with 64MB).
The only real problem is that the stage[123] compilers would be built
with -B./ or -Bstage[12]/ which is fine, and additionally -B${prefix},
which is miserably broken, considering that nothing is supposed to be
in ${prefix} yet. That one problem causes everything everywhere to
break, since things like ld.exe and crt0.o suddently can't be found. I
guess it's a bug in configure.
The one-stop-shopping solution is to
1) Mount \cygnus\cygwin-b20\H-yadda-yadda as /usr (which is what
I should have done originally; it took me forever to figure
out that H-* == usr).
2) Configure with --prefix=/usr --enable-languages=c++ (the others
seem to be impossible regardless of flags). This means that
the stage* compilers get called with -B/usr which will allow
the other backend tools to be found, since they already exist.
3) Bootstrap and install as normal. This will replace the shipped
version of egcs (1.0.something?) with whatever just got built.
Now if I can work around the different "nan" signatures in the math
library, I can get libstdc++-v3 built. Cool!
Phil
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: build successful (was Re: winsup in EGCS?)
1999-06-20 12:04 build successful (was Re: winsup in EGCS?) Phil Edwards
@ 1999-06-20 14:47 ` Gabriel Galibourg
1999-06-30 22:10 ` Gabriel Galibourg
1999-06-20 18:43 ` libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)] Mumit Khan
1999-06-30 22:10 ` build successful (was Re: winsup in EGCS?) Phil Edwards
2 siblings, 1 reply; 6+ messages in thread
From: Gabriel Galibourg @ 1999-06-20 14:47 UTC (permalink / raw)
To: cygwin; +Cc: Phil Edwards
Phil Edwards wrote:
>
> > $ <srcdir>/configure --prefix=/Cygnus/cygwin-b20 \
> > --exec-prefix=/Cygnus/cygwin-b20/H-i586-cygwin32 \
> > -v i586-cygwin32
> >
> > $ make bootstrap > make.log 2>&1
>
> I've been trying to use "make blah | tee /something 2&>1" but the error
> messages don't get captured in the logfile.
You should use:
make blah 2>&1 | tee /something
When using a pipe you need to redirect fd 2 into fd 1 BEFORE you use the pipe, if redirecting into a file then you need to redirect AFTER the redirection of fd 1 (into the file) ... it may look bizarre at first glance but when you really look into it it really all makes sense.
> snap .....
Cheers,
Gabriel.
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
* libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)]
1999-06-20 12:04 build successful (was Re: winsup in EGCS?) Phil Edwards
1999-06-20 14:47 ` Gabriel Galibourg
@ 1999-06-20 18:43 ` Mumit Khan
1999-06-30 22:10 ` Mumit Khan
1999-06-30 22:10 ` build successful (was Re: winsup in EGCS?) Phil Edwards
2 siblings, 1 reply; 6+ messages in thread
From: Mumit Khan @ 1999-06-20 18:43 UTC (permalink / raw)
To: cygwin
Phil Edwards <pedwards@jaj.com> writes:
>
> Now if I can work around the different "nan" signatures in the math
> library, I can get libstdc++-v3 built. Cool!
>
This is something from way back, but may help.
Thu Apr 20 09:49:58 1999 Mumit Khan <khan@xraylith.wisc.edu>
* bits/c++config.h (mbstate_t): Define for Cygwin as well.
* bits/std_cwchar.h: Cygwin declares conflicting prototypes.
* math/complex-stub.h (nan): Newlib's nan has different signature.
* math/mathconf.h (NAN): Define for Cygwin/newlib.
Index: bits/c++config.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/bits/c++config.h,v
retrieving revision 1.19
diff -u -3 -p -r1.19 c++config.h
--- bits/c++config.h 1999/03/18 23:26:19 1.19
+++ bits/c++config.h 1999/04/22 15:50:52
@@ -89,7 +89,7 @@ enum
#undef _ISbit
#endif
-#ifdef __sun__
+#if defined (__sun__) || defined (__CYGWIN__)
/* When we auto-generate this file this definition will done automatically
as well. */
typedef struct {
Index: bits/std_cwchar.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/bits/std_cwchar.h,v
retrieving revision 1.5
diff -u -3 -p -r1.5 std_cwchar.h
--- bits/std_cwchar.h 1998/12/31 23:15:39 1.5
+++ bits/std_cwchar.h 1999/04/22 15:50:52
@@ -38,6 +38,7 @@
# include_next <wchar.h>
+#ifndef __CYGWIN__
// XXX These don't belong here only if the system header does not define them.
extern "C" {
int wmemcmp (const wchar_t *__s1, const wchar_t *__s2, size_t __n);
@@ -47,5 +48,6 @@ wchar_t *wmemmove (wchar_t *__s1, const
wchar_t *wmemcpy (wchar_t *__s1, const wchar_t *__s2, size_t __n);
wchar_t *wmemset (wchar_t *__s, wchar_t __c, size_t __n);
}
+#endif
#endif
Index: math/complex-stub.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/math/complex-stub.h,v
retrieving revision 1.6
diff -u -3 -p -r1.6 complex-stub.h
--- math/complex-stub.h 1998/12/31 23:15:40 1.6
+++ math/complex-stub.h 1999/04/22 15:50:52
@@ -86,6 +86,8 @@ double __mycabs (__complex__ double x);
float __mycabsf (__complex__ float x);
long double __mycabsl (__complex__ long double x);
+#ifndef __CYGWIN__
double nan (const char *s);
+#endif
#endif
Index: math/mathconf.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/math/mathconf.h,v
retrieving revision 1.7
diff -u -3 -p -r1.7 mathconf.h
--- math/mathconf.h 1998/12/31 23:15:40 1.7
+++ math/mathconf.h 1999/04/22 15:50:52
@@ -113,8 +113,13 @@ typedef int Int32_t __attribute ((mode (
# endif
#endif
+
#ifndef NAN
+#ifdef __CYGWIN__
+# define NAN (nan ())
+#else
# define NAN (nan (""))
+#endif
#endif
Regards,
Mumit
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
* build successful (was Re: winsup in EGCS?)
1999-06-20 12:04 build successful (was Re: winsup in EGCS?) Phil Edwards
1999-06-20 14:47 ` Gabriel Galibourg
1999-06-20 18:43 ` libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)] Mumit Khan
@ 1999-06-30 22:10 ` Phil Edwards
2 siblings, 0 replies; 6+ messages in thread
From: Phil Edwards @ 1999-06-30 22:10 UTC (permalink / raw)
To: cygwin
> $ <srcdir>/configure --prefix=/Cygnus/cygwin-b20 \
> --exec-prefix=/Cygnus/cygwin-b20/H-i586-cygwin32 \
> -v i586-cygwin32
>
> $ make bootstrap > make.log 2>&1
I've been trying to use "make blah | tee /something 2&>1" but the error
messages don't get captured in the logfile.
> When it builds, just install it elsewhere:
>
> $ make prefix=/home/EGCS --exec-prefix=/home/EGCS/H-i586-cygwin32 \
> install
Oh. I've never tried to mess around with --exec-prefix. Never seems
to do what I thought it would, and since "uninstall" has never ever
worked, I don't experiment any more (egcs takes nearly six hours to build
on a P133 with 64MB).
The only real problem is that the stage[123] compilers would be built
with -B./ or -Bstage[12]/ which is fine, and additionally -B${prefix},
which is miserably broken, considering that nothing is supposed to be
in ${prefix} yet. That one problem causes everything everywhere to
break, since things like ld.exe and crt0.o suddently can't be found. I
guess it's a bug in configure.
The one-stop-shopping solution is to
1) Mount \cygnus\cygwin-b20\H-yadda-yadda as /usr (which is what
I should have done originally; it took me forever to figure
out that H-* == usr).
2) Configure with --prefix=/usr --enable-languages=c++ (the others
seem to be impossible regardless of flags). This means that
the stage* compilers get called with -B/usr which will allow
the other backend tools to be found, since they already exist.
3) Bootstrap and install as normal. This will replace the shipped
version of egcs (1.0.something?) with whatever just got built.
Now if I can work around the different "nan" signatures in the math
library, I can get libstdc++-v3 built. Cool!
Phil
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: build successful (was Re: winsup in EGCS?)
1999-06-20 14:47 ` Gabriel Galibourg
@ 1999-06-30 22:10 ` Gabriel Galibourg
0 siblings, 0 replies; 6+ messages in thread
From: Gabriel Galibourg @ 1999-06-30 22:10 UTC (permalink / raw)
To: cygwin; +Cc: Phil Edwards
Phil Edwards wrote:
>
> > $ <srcdir>/configure --prefix=/Cygnus/cygwin-b20 \
> > --exec-prefix=/Cygnus/cygwin-b20/H-i586-cygwin32 \
> > -v i586-cygwin32
> >
> > $ make bootstrap > make.log 2>&1
>
> I've been trying to use "make blah | tee /something 2&>1" but the error
> messages don't get captured in the logfile.
You should use:
make blah 2>&1 | tee /something
When using a pipe you need to redirect fd 2 into fd 1 BEFORE you use the pipe, if redirecting into a file then you need to redirect AFTER the redirection of fd 1 (into the file) ... it may look bizarre at first glance but when you really look into it it really all makes sense.
> snap .....
Cheers,
Gabriel.
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
* libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)]
1999-06-20 18:43 ` libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)] Mumit Khan
@ 1999-06-30 22:10 ` Mumit Khan
0 siblings, 0 replies; 6+ messages in thread
From: Mumit Khan @ 1999-06-30 22:10 UTC (permalink / raw)
To: cygwin
Phil Edwards <pedwards@jaj.com> writes:
>
> Now if I can work around the different "nan" signatures in the math
> library, I can get libstdc++-v3 built. Cool!
>
This is something from way back, but may help.
Thu Apr 20 09:49:58 1999 Mumit Khan <khan@xraylith.wisc.edu>
* bits/c++config.h (mbstate_t): Define for Cygwin as well.
* bits/std_cwchar.h: Cygwin declares conflicting prototypes.
* math/complex-stub.h (nan): Newlib's nan has different signature.
* math/mathconf.h (NAN): Define for Cygwin/newlib.
Index: bits/c++config.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/bits/c++config.h,v
retrieving revision 1.19
diff -u -3 -p -r1.19 c++config.h
--- bits/c++config.h 1999/03/18 23:26:19 1.19
+++ bits/c++config.h 1999/04/22 15:50:52
@@ -89,7 +89,7 @@ enum
#undef _ISbit
#endif
-#ifdef __sun__
+#if defined (__sun__) || defined (__CYGWIN__)
/* When we auto-generate this file this definition will done automatically
as well. */
typedef struct {
Index: bits/std_cwchar.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/bits/std_cwchar.h,v
retrieving revision 1.5
diff -u -3 -p -r1.5 std_cwchar.h
--- bits/std_cwchar.h 1998/12/31 23:15:39 1.5
+++ bits/std_cwchar.h 1999/04/22 15:50:52
@@ -38,6 +38,7 @@
# include_next <wchar.h>
+#ifndef __CYGWIN__
// XXX These don't belong here only if the system header does not define them.
extern "C" {
int wmemcmp (const wchar_t *__s1, const wchar_t *__s2, size_t __n);
@@ -47,5 +48,6 @@ wchar_t *wmemmove (wchar_t *__s1, const
wchar_t *wmemcpy (wchar_t *__s1, const wchar_t *__s2, size_t __n);
wchar_t *wmemset (wchar_t *__s, wchar_t __c, size_t __n);
}
+#endif
#endif
Index: math/complex-stub.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/math/complex-stub.h,v
retrieving revision 1.6
diff -u -3 -p -r1.6 complex-stub.h
--- math/complex-stub.h 1998/12/31 23:15:40 1.6
+++ math/complex-stub.h 1999/04/22 15:50:52
@@ -86,6 +86,8 @@ double __mycabs (__complex__ double x);
float __mycabsf (__complex__ float x);
long double __mycabsl (__complex__ long double x);
+#ifndef __CYGWIN__
double nan (const char *s);
+#endif
#endif
Index: math/mathconf.h
===================================================================
RCS file: /cvs/libstdc++/libstdc++/math/mathconf.h,v
retrieving revision 1.7
diff -u -3 -p -r1.7 mathconf.h
--- math/mathconf.h 1998/12/31 23:15:40 1.7
+++ math/mathconf.h 1999/04/22 15:50:52
@@ -113,8 +113,13 @@ typedef int Int32_t __attribute ((mode (
# endif
#endif
+
#ifndef NAN
+#ifdef __CYGWIN__
+# define NAN (nan ())
+#else
# define NAN (nan (""))
+#endif
#endif
Regards,
Mumit
--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~1999-06-30 22:10 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1999-06-20 12:04 build successful (was Re: winsup in EGCS?) Phil Edwards
1999-06-20 14:47 ` Gabriel Galibourg
1999-06-30 22:10 ` Gabriel Galibourg
1999-06-20 18:43 ` libstdc++-v3 patch [Re: build successful (was Re: winsup in EGCS?)] Mumit Khan
1999-06-30 22:10 ` Mumit Khan
1999-06-30 22:10 ` build successful (was Re: winsup in EGCS?) Phil Edwards
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).