* Announcing libstdc++-2.90.7
@ 1999-12-21 21:15 Benjamin Kosnik
1999-12-22 7:38 ` Matthew Brown
1999-12-31 23:54 ` Benjamin Kosnik
0 siblings, 2 replies; 16+ messages in thread
From: Benjamin Kosnik @ 1999-12-21 21:15 UTC (permalink / raw)
To: libstdc++, egcs, info-gnu
1999-12-21
Release Notes
-------------
The Standard C++ Library v3, or libstc++-2.90.x, is an ongoing
project to implement the ISO 14882 Standard C++ library as described
in chapters 17 through 27 and annex D, as a drop-in replacement
for the current (ARM-conformant) library.
This is the eigth snapshot of the libstdc++ rewrite. It is still
incomplet and incorrekt.
The Standard C++ Library v3, or libstc++-2.90.x, follows an open
development model, attempting to be fully buzzword, bazaar, and GNU
compliant. Full details on participating, including contributor
guidelines, mailing list subscription, mailing list archives,
up-to-date documentation, and various and sundry other details can be
found at the following URL:
http://sourceware.cygnus.com/libstdc++/
New:
---
- Now compiling with -Wall -Werror -Winline works, and is on by default.
- Conditional compilation for hosts that don't support wchar_t.
- Attempts have been made to get this to compile hosted on cygwin.
Mumit Khan has done bunches here.
- Chapter 22:
Kevin Ediger did some killer formatting work for the floating point parts of
num_put::do_put.
- Chapter 26:
valarray fixes:
the non-standard member function
'valarray<T>::product()' is removed for Standard compliance purpose.
Some missing non member functions (e.g. tanh) are added.
valarray::sum() now uses operator+= instead of operator+.
valarray::min() and valarray::max() no longer causes parse errors.
Typographical errors that used to cause a seg fault are corrected.
complex fixes: some missing constructors are added.
Optimized calling of cmath funcs for targets that support builtin functions.
mknumeric_limits was robustified.
- Chapter 27:
Preliminary inlining for all buffering and iostream classes and manipulators.
Testing for istream and ostream operators complete.
istream validation testing complete.
ostream formatting validations mostly complete.
ostream<<operator(float, double, etc.) now work.
Testing for correct stream state and thrown exceptions complete.
A lot of stringbuf changes.
Attempts to make includes take as little time as possible.
streambuf to streambuf copies now work.
A brain-dead and in-elegant (but conformant) facet caching mechanism
replaces the elegant but non-conformant design.
iostream callbacks now work correctly.
A lot of the library working group issues are now implemented.
- Additional documentation by Phil Edwards.
- New website and mailing list interface.
- For CVS egcs, -O2 and tree-level inlining works.
- Many, many bug fixes.
More info is available on the project website, URL as above.
-Benjamin
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-21 21:15 Announcing libstdc++-2.90.7 Benjamin Kosnik
@ 1999-12-22 7:38 ` Matthew Brown
1999-12-22 11:15 ` Mumit Khan
1999-12-31 23:54 ` Matthew Brown
1999-12-31 23:54 ` Benjamin Kosnik
1 sibling, 2 replies; 16+ messages in thread
From: Matthew Brown @ 1999-12-22 7:38 UTC (permalink / raw)
To: Benjamin Kosnik, libstdc++, egcs, info-gnu
I just attempted a compile and installation. All goes well (using Mumit's
cygwin limits) until the installation. The makefile seems to desire a file
in the 'bits' source directory by the name 'ctype_base.h'. This file seems
to be somewhat platform-specific and appears, at first look, to be copied
from the 'config' directory during configuration time; however, this is not
happening for me. Here is my configuration:
- Cygwin b20.1
- configure run with '--prefix=<cygwin
dir> --enable-shared --enable-static --enable-namespaces'
For now, I will copy the default version of the file and see if things work.
I know that this is perilously close to the holidays, but if anyone is still
watching who has some input, please pass it on.
Thanks,
-- Matthew Brown
----- Original Message -----
From: Benjamin Kosnik <bkoz@cygnus.com>
To: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>;
<info-gnu@gnu.org>
Sent: Wednesday, December 22, 1999 12:15 AM
Subject: Announcing libstdc++-2.90.7
>
> 1999-12-21
>
> Release Notes
> -------------
> The Standard C++ Library v3, or libstc++-2.90.x, is an ongoing
> project to implement the ISO 14882 Standard C++ library as described
> in chapters 17 through 27 and annex D, as a drop-in replacement
> for the current (ARM-conformant) library.
>
> This is the eigth snapshot of the libstdc++ rewrite. It is still
> incomplet and incorrekt.
>
> The Standard C++ Library v3, or libstc++-2.90.x, follows an open
> development model, attempting to be fully buzzword, bazaar, and GNU
> compliant. Full details on participating, including contributor
> guidelines, mailing list subscription, mailing list archives,
> up-to-date documentation, and various and sundry other details can be
> found at the following URL:
>
> http://sourceware.cygnus.com/libstdc++/
>
>
> New:
> ---
> - Now compiling with -Wall -Werror -Winline works, and is on by default.
> - Conditional compilation for hosts that don't support wchar_t.
> - Attempts have been made to get this to compile hosted on cygwin.
> Mumit Khan has done bunches here.
> - Chapter 22:
> Kevin Ediger did some killer formatting work for the floating point
parts of
> num_put::do_put.
> - Chapter 26:
> valarray fixes:
> the non-standard member function
> 'valarray<T>::product()' is removed for Standard compliance purpose.
> Some missing non member functions (e.g. tanh) are added.
> valarray::sum() now uses operator+= instead of operator+.
> valarray::min() and valarray::max() no longer causes parse errors.
> Typographical errors that used to cause a seg fault are corrected.
> complex fixes: some missing constructors are added.
> Optimized calling of cmath funcs for targets that support builtin
functions.
> mknumeric_limits was robustified.
> - Chapter 27:
> Preliminary inlining for all buffering and iostream classes and
manipulators.
> Testing for istream and ostream operators complete.
> istream validation testing complete.
> ostream formatting validations mostly complete.
> ostream<<operator(float, double, etc.) now work.
> Testing for correct stream state and thrown exceptions complete.
> A lot of stringbuf changes.
> Attempts to make includes take as little time as possible.
> streambuf to streambuf copies now work.
> A brain-dead and in-elegant (but conformant) facet caching mechanism
> replaces the elegant but non-conformant design.
> iostream callbacks now work correctly.
> A lot of the library working group issues are now implemented.
> - Additional documentation by Phil Edwards.
> - New website and mailing list interface.
> - For CVS egcs, -O2 and tree-level inlining works.
> - Many, many bug fixes.
>
> More info is available on the project website, URL as above.
>
> -Benjamin
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 7:38 ` Matthew Brown
@ 1999-12-22 11:15 ` Mumit Khan
1999-12-22 11:55 ` Matthew Brown
` (3 more replies)
1999-12-31 23:54 ` Matthew Brown
1 sibling, 4 replies; 16+ messages in thread
From: Mumit Khan @ 1999-12-22 11:15 UTC (permalink / raw)
To: Matthew Brown; +Cc: libstdc++, egcs
On Wed, 22 Dec 1999, Matthew Brown wrote:
> I just attempted a compile and installation. All goes well (using Mumit's
> cygwin limits) until the installation. The makefile seems to desire a file
> in the 'bits' source directory by the name 'ctype_base.h'. This file seems
> to be somewhat platform-specific and appears, at first look, to be copied
> from the 'config' directory during configuration time; however, this is not
> happening for me. Here is my configuration:
>
> - Cygwin b20.1
> - configure run with '--prefix=<cygwin
> dir> --enable-shared --enable-static --enable-namespaces'
Don't know what the problem is, but I could cross-build on Linux just
fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
libgcc with -fhonor-std, but I assume you know that.
You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
The package contains a rebuilt libgcc.a as well.
Regards,
Mumit
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 11:15 ` Mumit Khan
@ 1999-12-22 11:55 ` Matthew Brown
1999-12-31 23:54 ` Matthew Brown
1999-12-22 22:53 ` Benjamin Kosnik
` (2 subsequent siblings)
3 siblings, 1 reply; 16+ messages in thread
From: Matthew Brown @ 1999-12-22 11:55 UTC (permalink / raw)
To: Mumit Khan; +Cc: libstdc++, egcs
Just guessing here but there is a srcdir/config/bits/linux whereas there is
no srcdir/config/bits/cygwin. I took the ctypes_base.h from
srcdir/config/bits/default. I don't know if this will work because I have
not finished my porting project yet -- so far it compiles, but I don't know
about linking or running.
Thanks. I'm out of here 'till next week.
-- Matthew Brown
----- Original Message -----
From: Mumit Khan <khan@NanoTech.Wisc.EDU>
To: Matthew Brown <mbrown@mediadb.net>
Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
Sent: Wednesday, December 22, 1999 2:15 PM
Subject: Re: Announcing libstdc++-2.90.7
> On Wed, 22 Dec 1999, Matthew Brown wrote:
>
> > I just attempted a compile and installation. All goes well (using
Mumit's
> > cygwin limits) until the installation. The makefile seems to desire a
file
> > in the 'bits' source directory by the name 'ctype_base.h'. This file
seems
> > to be somewhat platform-specific and appears, at first look, to be
copied
> > from the 'config' directory during configuration time; however, this is
not
> > happening for me. Here is my configuration:
> >
> > - Cygwin b20.1
> > - configure run with '--prefix=<cygwin
> > dir> --enable-shared --enable-static --enable-namespaces'
>
> Don't know what the problem is, but I could cross-build on Linux just
> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> libgcc with -fhonor-std, but I assume you know that.
>
> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
>
> The package contains a rebuilt libgcc.a as well.
>
> Regards,
> Mumit
>
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 11:15 ` Mumit Khan
1999-12-22 11:55 ` Matthew Brown
@ 1999-12-22 22:53 ` Benjamin Kosnik
1999-12-31 23:54 ` Benjamin Kosnik
1999-12-23 3:05 ` Re[2]: " Roman Belenov
1999-12-31 23:54 ` Mumit Khan
3 siblings, 1 reply; 16+ messages in thread
From: Benjamin Kosnik @ 1999-12-22 22:53 UTC (permalink / raw)
To: Mumit Khan; +Cc: Matthew Brown, libstdc++, egcs
> > - Cygwin b20.1
> > - configure run with '--prefix=<cygwin
> > dir> --enable-shared --enable-static --enable-namespaces'
>
> Don't know what the problem is, but I could cross-build on Linux just
> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> libgcc with -fhonor-std, but I assume you know that.
how about those --enable flags!!!!!!!! I'm psyched. Kudos to phil.
> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
can we add this to the downloads page please? Does this mean that the
signal handling in cygwin can actually run mknumeric_limits now? Or is
that still in an odd state?
-benjamin
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re[2]: Announcing libstdc++-2.90.7
1999-12-22 11:15 ` Mumit Khan
1999-12-22 11:55 ` Matthew Brown
1999-12-22 22:53 ` Benjamin Kosnik
@ 1999-12-23 3:05 ` Roman Belenov
1999-12-27 9:12 ` Matthew Brown
1999-12-31 23:54 ` Roman Belenov
1999-12-31 23:54 ` Mumit Khan
3 siblings, 2 replies; 16+ messages in thread
From: Roman Belenov @ 1999-12-23 3:05 UTC (permalink / raw)
To: Mumit Khan; +Cc: libstdc++, egcs
Hello Mumit,
It seems that there are problems when building in source directory
instead of making a new directory and running configure from where. I
had the same problem as Matt under Linux; building in other directory
(and installing from there) worked without problems.
MK> On Wed, 22 Dec 1999, Matthew Brown wrote:
>> I just attempted a compile and installation. All goes well (using Mumit's
>> cygwin limits) until the installation. The makefile seems to desire a file
>> in the 'bits' source directory by the name 'ctype_base.h'. This file seems
>> to be somewhat platform-specific and appears, at first look, to be copied
>> from the 'config' directory during configuration time; however, this is not
>> happening for me. Here is my configuration:
>>
>> - Cygwin b20.1
>> - configure run with '--prefix=<cygwin
>> dir> --enable-shared --enable-static --enable-namespaces'
MK> Don't know what the problem is, but I could cross-build on Linux just
MK> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
MK> libgcc with -fhonor-std, but I assume you know that.
MK> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
MK> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
MK> The package contains a rebuilt libgcc.a as well.
MK> Regards,
MK> Mumit
Best regards,
Roman mailto:roman@nstl.nnov.ru
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Re[2]: Announcing libstdc++-2.90.7
1999-12-23 3:05 ` Re[2]: " Roman Belenov
@ 1999-12-27 9:12 ` Matthew Brown
1999-12-27 9:37 ` Re[4]: " Roman Belenov
1999-12-31 23:54 ` Re[2]: " Matthew Brown
1999-12-31 23:54 ` Roman Belenov
1 sibling, 2 replies; 16+ messages in thread
From: Matthew Brown @ 1999-12-27 9:12 UTC (permalink / raw)
To: Roman Belenov, Benjamin Kosnik; +Cc: libstdc++, egcs
I *did* configure and build from a separate directory -- it's a habit with
me.
I have successfully built and installed on my Solaris box, but it, again,
required a tweak. This one was with <srcdir>/config/solaris/ctype.cc:
The gcc 2.95.1 compiler complains about '__trans_upper' and __trans_lower'
being 'int *' while the left side of the related assignments are 'long int
*'. Performing a C-style cast to 'long int *' in these four statements
allows the compile to succeed.
-- Matthew Brown
----- Original Message -----
From: Roman Belenov <roman@nstl.nnov.ru>
To: Mumit Khan <khan@NanoTech.Wisc.EDU>
Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
Sent: Thursday, December 23, 1999 6:05 AM
Subject: Re[2]: Announcing libstdc++-2.90.7
> Hello Mumit,
>
> It seems that there are problems when building in source directory
> instead of making a new directory and running configure from where. I
> had the same problem as Matt under Linux; building in other directory
> (and installing from there) worked without problems.
>
> MK> On Wed, 22 Dec 1999, Matthew Brown wrote:
>
> >> I just attempted a compile and installation. All goes well (using
Mumit's
> >> cygwin limits) until the installation. The makefile seems to desire a
file
> >> in the 'bits' source directory by the name 'ctype_base.h'. This file
seems
> >> to be somewhat platform-specific and appears, at first look, to be
copied
> >> from the 'config' directory during configuration time; however, this is
not
> >> happening for me. Here is my configuration:
> >>
> >> - Cygwin b20.1
> >> - configure run with '--prefix=<cygwin
> >> dir> --enable-shared --enable-static --enable-namespaces'
>
> MK> Don't know what the problem is, but I could cross-build on Linux just
> MK> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> MK> libgcc with -fhonor-std, but I assume you know that.
>
> MK> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> MK>
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
>
> MK> The package contains a rebuilt libgcc.a as well.
>
> MK> Regards,
> MK> Mumit
>
>
>
>
> Best regards,
> Roman mailto:roman@nstl.nnov.ru
>
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re[4]: Announcing libstdc++-2.90.7
1999-12-27 9:12 ` Matthew Brown
@ 1999-12-27 9:37 ` Roman Belenov
1999-12-31 23:54 ` Roman Belenov
1999-12-31 23:54 ` Re[2]: " Matthew Brown
1 sibling, 1 reply; 16+ messages in thread
From: Roman Belenov @ 1999-12-27 9:37 UTC (permalink / raw)
To: Matthew Brown; +Cc: libstdc++, egcs
Hello Matthew,
Another build attempt revealed that there are two different problems -
on Linux I succesfully built libstdc++ in separate directory and got
error on install stage when building in source dir; on NT system the
error appeared no matter where the build directory was.
MB> I *did* configure and build from a separate directory -- it's a habit with
MB> me.
MB> I have successfully built and installed on my Solaris box, but it, again,
MB> required a tweak. This one was with <srcdir>/config/solaris/ctype.cc:
MB> The gcc 2.95.1 compiler complains about '__trans_upper' and __trans_lower'
MB> being 'int *' while the left side of the related assignments are 'long int
MB> *'. Performing a C-style cast to 'long int *' in these four statements
MB> allows the compile to succeed.
MB> -- Matthew Brown
MB> ----- Original Message -----
MB> From: Roman Belenov <roman@nstl.nnov.ru>
MB> To: Mumit Khan <khan@NanoTech.Wisc.EDU>
MB> Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
MB> Sent: Thursday, December 23, 1999 6:05 AM
MB> Subject: Re[2]: Announcing libstdc++-2.90.7
>> Hello Mumit,
>>
>> It seems that there are problems when building in source directory
>> instead of making a new directory and running configure from where. I
>> had the same problem as Matt under Linux; building in other directory
>> (and installing from there) worked without problems.
>>
--------------- skipped --------------------------
Best regards,
Roman mailto:roman@nstl.nnov.ru
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Re[2]: Announcing libstdc++-2.90.7
1999-12-27 9:12 ` Matthew Brown
1999-12-27 9:37 ` Re[4]: " Roman Belenov
@ 1999-12-31 23:54 ` Matthew Brown
1 sibling, 0 replies; 16+ messages in thread
From: Matthew Brown @ 1999-12-31 23:54 UTC (permalink / raw)
To: Roman Belenov, Benjamin Kosnik; +Cc: libstdc++, egcs
I *did* configure and build from a separate directory -- it's a habit with
me.
I have successfully built and installed on my Solaris box, but it, again,
required a tweak. This one was with <srcdir>/config/solaris/ctype.cc:
The gcc 2.95.1 compiler complains about '__trans_upper' and __trans_lower'
being 'int *' while the left side of the related assignments are 'long int
*'. Performing a C-style cast to 'long int *' in these four statements
allows the compile to succeed.
-- Matthew Brown
----- Original Message -----
From: Roman Belenov <roman@nstl.nnov.ru>
To: Mumit Khan <khan@NanoTech.Wisc.EDU>
Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
Sent: Thursday, December 23, 1999 6:05 AM
Subject: Re[2]: Announcing libstdc++-2.90.7
> Hello Mumit,
>
> It seems that there are problems when building in source directory
> instead of making a new directory and running configure from where. I
> had the same problem as Matt under Linux; building in other directory
> (and installing from there) worked without problems.
>
> MK> On Wed, 22 Dec 1999, Matthew Brown wrote:
>
> >> I just attempted a compile and installation. All goes well (using
Mumit's
> >> cygwin limits) until the installation. The makefile seems to desire a
file
> >> in the 'bits' source directory by the name 'ctype_base.h'. This file
seems
> >> to be somewhat platform-specific and appears, at first look, to be
copied
> >> from the 'config' directory during configuration time; however, this is
not
> >> happening for me. Here is my configuration:
> >>
> >> - Cygwin b20.1
> >> - configure run with '--prefix=<cygwin
> >> dir> --enable-shared --enable-static --enable-namespaces'
>
> MK> Don't know what the problem is, but I could cross-build on Linux just
> MK> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> MK> libgcc with -fhonor-std, but I assume you know that.
>
> MK> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> MK>
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
>
> MK> The package contains a rebuilt libgcc.a as well.
>
> MK> Regards,
> MK> Mumit
>
>
>
>
> Best regards,
> Roman mailto:roman@nstl.nnov.ru
>
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 11:15 ` Mumit Khan
` (2 preceding siblings ...)
1999-12-23 3:05 ` Re[2]: " Roman Belenov
@ 1999-12-31 23:54 ` Mumit Khan
3 siblings, 0 replies; 16+ messages in thread
From: Mumit Khan @ 1999-12-31 23:54 UTC (permalink / raw)
To: Matthew Brown; +Cc: libstdc++, egcs
On Wed, 22 Dec 1999, Matthew Brown wrote:
> I just attempted a compile and installation. All goes well (using Mumit's
> cygwin limits) until the installation. The makefile seems to desire a file
> in the 'bits' source directory by the name 'ctype_base.h'. This file seems
> to be somewhat platform-specific and appears, at first look, to be copied
> from the 'config' directory during configuration time; however, this is not
> happening for me. Here is my configuration:
>
> - Cygwin b20.1
> - configure run with '--prefix=<cygwin
> dir> --enable-shared --enable-static --enable-namespaces'
Don't know what the problem is, but I could cross-build on Linux just
fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
libgcc with -fhonor-std, but I assume you know that.
You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
The package contains a rebuilt libgcc.a as well.
Regards,
Mumit
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re[4]: Announcing libstdc++-2.90.7
1999-12-27 9:37 ` Re[4]: " Roman Belenov
@ 1999-12-31 23:54 ` Roman Belenov
0 siblings, 0 replies; 16+ messages in thread
From: Roman Belenov @ 1999-12-31 23:54 UTC (permalink / raw)
To: Matthew Brown; +Cc: libstdc++, egcs
Hello Matthew,
Another build attempt revealed that there are two different problems -
on Linux I succesfully built libstdc++ in separate directory and got
error on install stage when building in source dir; on NT system the
error appeared no matter where the build directory was.
MB> I *did* configure and build from a separate directory -- it's a habit with
MB> me.
MB> I have successfully built and installed on my Solaris box, but it, again,
MB> required a tweak. This one was with <srcdir>/config/solaris/ctype.cc:
MB> The gcc 2.95.1 compiler complains about '__trans_upper' and __trans_lower'
MB> being 'int *' while the left side of the related assignments are 'long int
MB> *'. Performing a C-style cast to 'long int *' in these four statements
MB> allows the compile to succeed.
MB> -- Matthew Brown
MB> ----- Original Message -----
MB> From: Roman Belenov <roman@nstl.nnov.ru>
MB> To: Mumit Khan <khan@NanoTech.Wisc.EDU>
MB> Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
MB> Sent: Thursday, December 23, 1999 6:05 AM
MB> Subject: Re[2]: Announcing libstdc++-2.90.7
>> Hello Mumit,
>>
>> It seems that there are problems when building in source directory
>> instead of making a new directory and running configure from where. I
>> had the same problem as Matt under Linux; building in other directory
>> (and installing from there) worked without problems.
>>
--------------- skipped --------------------------
Best regards,
Roman mailto:roman@nstl.nnov.ru
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 22:53 ` Benjamin Kosnik
@ 1999-12-31 23:54 ` Benjamin Kosnik
0 siblings, 0 replies; 16+ messages in thread
From: Benjamin Kosnik @ 1999-12-31 23:54 UTC (permalink / raw)
To: Mumit Khan; +Cc: Matthew Brown, libstdc++, egcs
> > - Cygwin b20.1
> > - configure run with '--prefix=<cygwin
> > dir> --enable-shared --enable-static --enable-namespaces'
>
> Don't know what the problem is, but I could cross-build on Linux just
> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> libgcc with -fhonor-std, but I assume you know that.
how about those --enable flags!!!!!!!! I'm psyched. Kudos to phil.
> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
can we add this to the downloads page please? Does this mean that the
signal handling in cygwin can actually run mknumeric_limits now? Or is
that still in an odd state?
-benjamin
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 7:38 ` Matthew Brown
1999-12-22 11:15 ` Mumit Khan
@ 1999-12-31 23:54 ` Matthew Brown
1 sibling, 0 replies; 16+ messages in thread
From: Matthew Brown @ 1999-12-31 23:54 UTC (permalink / raw)
To: Benjamin Kosnik, libstdc++, egcs, info-gnu
I just attempted a compile and installation. All goes well (using Mumit's
cygwin limits) until the installation. The makefile seems to desire a file
in the 'bits' source directory by the name 'ctype_base.h'. This file seems
to be somewhat platform-specific and appears, at first look, to be copied
from the 'config' directory during configuration time; however, this is not
happening for me. Here is my configuration:
- Cygwin b20.1
- configure run with '--prefix=<cygwin
dir> --enable-shared --enable-static --enable-namespaces'
For now, I will copy the default version of the file and see if things work.
I know that this is perilously close to the holidays, but if anyone is still
watching who has some input, please pass it on.
Thanks,
-- Matthew Brown
----- Original Message -----
From: Benjamin Kosnik <bkoz@cygnus.com>
To: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>;
<info-gnu@gnu.org>
Sent: Wednesday, December 22, 1999 12:15 AM
Subject: Announcing libstdc++-2.90.7
>
> 1999-12-21
>
> Release Notes
> -------------
> The Standard C++ Library v3, or libstc++-2.90.x, is an ongoing
> project to implement the ISO 14882 Standard C++ library as described
> in chapters 17 through 27 and annex D, as a drop-in replacement
> for the current (ARM-conformant) library.
>
> This is the eigth snapshot of the libstdc++ rewrite. It is still
> incomplet and incorrekt.
>
> The Standard C++ Library v3, or libstc++-2.90.x, follows an open
> development model, attempting to be fully buzzword, bazaar, and GNU
> compliant. Full details on participating, including contributor
> guidelines, mailing list subscription, mailing list archives,
> up-to-date documentation, and various and sundry other details can be
> found at the following URL:
>
> http://sourceware.cygnus.com/libstdc++/
>
>
> New:
> ---
> - Now compiling with -Wall -Werror -Winline works, and is on by default.
> - Conditional compilation for hosts that don't support wchar_t.
> - Attempts have been made to get this to compile hosted on cygwin.
> Mumit Khan has done bunches here.
> - Chapter 22:
> Kevin Ediger did some killer formatting work for the floating point
parts of
> num_put::do_put.
> - Chapter 26:
> valarray fixes:
> the non-standard member function
> 'valarray<T>::product()' is removed for Standard compliance purpose.
> Some missing non member functions (e.g. tanh) are added.
> valarray::sum() now uses operator+= instead of operator+.
> valarray::min() and valarray::max() no longer causes parse errors.
> Typographical errors that used to cause a seg fault are corrected.
> complex fixes: some missing constructors are added.
> Optimized calling of cmath funcs for targets that support builtin
functions.
> mknumeric_limits was robustified.
> - Chapter 27:
> Preliminary inlining for all buffering and iostream classes and
manipulators.
> Testing for istream and ostream operators complete.
> istream validation testing complete.
> ostream formatting validations mostly complete.
> ostream<<operator(float, double, etc.) now work.
> Testing for correct stream state and thrown exceptions complete.
> A lot of stringbuf changes.
> Attempts to make includes take as little time as possible.
> streambuf to streambuf copies now work.
> A brain-dead and in-elegant (but conformant) facet caching mechanism
> replaces the elegant but non-conformant design.
> iostream callbacks now work correctly.
> A lot of the library working group issues are now implemented.
> - Additional documentation by Phil Edwards.
> - New website and mailing list interface.
> - For CVS egcs, -O2 and tree-level inlining works.
> - Many, many bug fixes.
>
> More info is available on the project website, URL as above.
>
> -Benjamin
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re: Announcing libstdc++-2.90.7
1999-12-22 11:55 ` Matthew Brown
@ 1999-12-31 23:54 ` Matthew Brown
0 siblings, 0 replies; 16+ messages in thread
From: Matthew Brown @ 1999-12-31 23:54 UTC (permalink / raw)
To: Mumit Khan; +Cc: libstdc++, egcs
Just guessing here but there is a srcdir/config/bits/linux whereas there is
no srcdir/config/bits/cygwin. I took the ctypes_base.h from
srcdir/config/bits/default. I don't know if this will work because I have
not finished my porting project yet -- so far it compiles, but I don't know
about linking or running.
Thanks. I'm out of here 'till next week.
-- Matthew Brown
----- Original Message -----
From: Mumit Khan <khan@NanoTech.Wisc.EDU>
To: Matthew Brown <mbrown@mediadb.net>
Cc: <libstdc++@sourceware.cygnus.com>; <egcs@egcs.cygnus.com>
Sent: Wednesday, December 22, 1999 2:15 PM
Subject: Re: Announcing libstdc++-2.90.7
> On Wed, 22 Dec 1999, Matthew Brown wrote:
>
> > I just attempted a compile and installation. All goes well (using
Mumit's
> > cygwin limits) until the installation. The makefile seems to desire a
file
> > in the 'bits' source directory by the name 'ctype_base.h'. This file
seems
> > to be somewhat platform-specific and appears, at first look, to be
copied
> > from the 'config' directory during configuration time; however, this is
not
> > happening for me. Here is my configuration:
> >
> > - Cygwin b20.1
> > - configure run with '--prefix=<cygwin
> > dir> --enable-shared --enable-static --enable-namespaces'
>
> Don't know what the problem is, but I could cross-build on Linux just
> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
> libgcc with -fhonor-std, but I assume you know that.
>
> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
>
> The package contains a rebuilt libgcc.a as well.
>
> Regards,
> Mumit
>
>
>
^ permalink raw reply [flat|nested] 16+ messages in thread
* Re[2]: Announcing libstdc++-2.90.7
1999-12-23 3:05 ` Re[2]: " Roman Belenov
1999-12-27 9:12 ` Matthew Brown
@ 1999-12-31 23:54 ` Roman Belenov
1 sibling, 0 replies; 16+ messages in thread
From: Roman Belenov @ 1999-12-31 23:54 UTC (permalink / raw)
To: Mumit Khan; +Cc: libstdc++, egcs
Hello Mumit,
It seems that there are problems when building in source directory
instead of making a new directory and running configure from where. I
had the same problem as Matt under Linux; building in other directory
(and installing from there) worked without problems.
MK> On Wed, 22 Dec 1999, Matthew Brown wrote:
>> I just attempted a compile and installation. All goes well (using Mumit's
>> cygwin limits) until the installation. The makefile seems to desire a file
>> in the 'bits' source directory by the name 'ctype_base.h'. This file seems
>> to be somewhat platform-specific and appears, at first look, to be copied
>> from the 'config' directory during configuration time; however, this is not
>> happening for me. Here is my configuration:
>>
>> - Cygwin b20.1
>> - configure run with '--prefix=<cygwin
>> dir> --enable-shared --enable-static --enable-namespaces'
MK> Don't know what the problem is, but I could cross-build on Linux just
MK> fine. BTW, if you use --enable-namespaces, you'll also have to rebuild
MK> libgcc with -fhonor-std, but I assume you know that.
MK> You can pick up prebuilt Cygwin and Mingw 2.90.7 at:
MK> http://www.xraylith.wisc.edu/~khan/software/gnu-win32/libstdc++-v3.html
MK> The package contains a rebuilt libgcc.a as well.
MK> Regards,
MK> Mumit
Best regards,
Roman mailto:roman@nstl.nnov.ru
^ permalink raw reply [flat|nested] 16+ messages in thread
* Announcing libstdc++-2.90.7
1999-12-21 21:15 Announcing libstdc++-2.90.7 Benjamin Kosnik
1999-12-22 7:38 ` Matthew Brown
@ 1999-12-31 23:54 ` Benjamin Kosnik
1 sibling, 0 replies; 16+ messages in thread
From: Benjamin Kosnik @ 1999-12-31 23:54 UTC (permalink / raw)
To: libstdc++, egcs, info-gnu
1999-12-21
Release Notes
-------------
The Standard C++ Library v3, or libstc++-2.90.x, is an ongoing
project to implement the ISO 14882 Standard C++ library as described
in chapters 17 through 27 and annex D, as a drop-in replacement
for the current (ARM-conformant) library.
This is the eigth snapshot of the libstdc++ rewrite. It is still
incomplet and incorrekt.
The Standard C++ Library v3, or libstc++-2.90.x, follows an open
development model, attempting to be fully buzzword, bazaar, and GNU
compliant. Full details on participating, including contributor
guidelines, mailing list subscription, mailing list archives,
up-to-date documentation, and various and sundry other details can be
found at the following URL:
http://sourceware.cygnus.com/libstdc++/
New:
---
- Now compiling with -Wall -Werror -Winline works, and is on by default.
- Conditional compilation for hosts that don't support wchar_t.
- Attempts have been made to get this to compile hosted on cygwin.
Mumit Khan has done bunches here.
- Chapter 22:
Kevin Ediger did some killer formatting work for the floating point parts of
num_put::do_put.
- Chapter 26:
valarray fixes:
the non-standard member function
'valarray<T>::product()' is removed for Standard compliance purpose.
Some missing non member functions (e.g. tanh) are added.
valarray::sum() now uses operator+= instead of operator+.
valarray::min() and valarray::max() no longer causes parse errors.
Typographical errors that used to cause a seg fault are corrected.
complex fixes: some missing constructors are added.
Optimized calling of cmath funcs for targets that support builtin functions.
mknumeric_limits was robustified.
- Chapter 27:
Preliminary inlining for all buffering and iostream classes and manipulators.
Testing for istream and ostream operators complete.
istream validation testing complete.
ostream formatting validations mostly complete.
ostream<<operator(float, double, etc.) now work.
Testing for correct stream state and thrown exceptions complete.
A lot of stringbuf changes.
Attempts to make includes take as little time as possible.
streambuf to streambuf copies now work.
A brain-dead and in-elegant (but conformant) facet caching mechanism
replaces the elegant but non-conformant design.
iostream callbacks now work correctly.
A lot of the library working group issues are now implemented.
- Additional documentation by Phil Edwards.
- New website and mailing list interface.
- For CVS egcs, -O2 and tree-level inlining works.
- Many, many bug fixes.
More info is available on the project website, URL as above.
-Benjamin
^ permalink raw reply [flat|nested] 16+ messages in thread
end of thread, other threads:[~1999-12-31 23:54 UTC | newest]
Thread overview: 16+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1999-12-21 21:15 Announcing libstdc++-2.90.7 Benjamin Kosnik
1999-12-22 7:38 ` Matthew Brown
1999-12-22 11:15 ` Mumit Khan
1999-12-22 11:55 ` Matthew Brown
1999-12-31 23:54 ` Matthew Brown
1999-12-22 22:53 ` Benjamin Kosnik
1999-12-31 23:54 ` Benjamin Kosnik
1999-12-23 3:05 ` Re[2]: " Roman Belenov
1999-12-27 9:12 ` Matthew Brown
1999-12-27 9:37 ` Re[4]: " Roman Belenov
1999-12-31 23:54 ` Roman Belenov
1999-12-31 23:54 ` Re[2]: " Matthew Brown
1999-12-31 23:54 ` Roman Belenov
1999-12-31 23:54 ` Mumit Khan
1999-12-31 23:54 ` Matthew Brown
1999-12-31 23:54 ` Benjamin Kosnik
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).