Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

I've just uploaded a test version of gcc 3.1.1 based on the main gcc
3.1.1 branch.

Danny Smith has put a lot of work into this release and currently has an
older version of gcc 3.1 available for testing in mingw.  This 3.1.1
version incorporates more bug fixes from him as well as from normal gcc
folk.

Danny reports that exception handling should work ok in this release
although I believe that there are still issues with exception handling
in dlls.

David Billinghurst has also contributed some changes to this release.

My main contributions are some configury reorg and of course, my old
friend, "-mno-cygwin".  In theory, "-mno-cygwin" should work much
better in this release.  It should even be functional for C++.

To that end, I'm also releasing an auxiliary package called
gcc-mingw-3_1-20020516-1.  This package contains header and library
files from the mingw release from the main mingw site, repackaged with
only the bits that are required for the cygwin version of gcc.  I
hope to be updating this as necessary (or getting someone to volunteer
to do this once things have settled down).

(Hi Egor)

So, for a fully capable version of gcc 3.1.1, you'll need to install the
test versions of both gcc and gcc-mingw.  Eventually, when gcc 3.1.1 is
no longer in testing, I'll add a dependency in gcc for gcc-mingw, so
that it will be pulled in automatically.

That brings us to gcc2-2.95.3-6.  As you might guess, this package is
just a repackaging of the old gcc-2.95.3 release.  The difference here,
though, is that all of the appropriate files are labelled gcc-2.exe
g++-2.exe gcc-2.man, etc.  The reason for this is to allow you to use
both the old and new versions of gcc on your system at the same time.
Then, when you're satisfied that you no longer require gcc2 stuff you
can uninstall the gcc2 package without affecting the gcc 3.1.1 stuff.
Or, you can just keep both around.

gcc 3.1.1 has been built with thread support enabled.  It also includes
the java compiler but I haven't actually verified that either works.
Theoretically, this could also include ada support, but if I do decide
to make this available, it will probably be as a separate package.  I
may do this with java eventually, too.

gcc 2.95.3-6 was built with (I hope) all of the recent
internationalization libraries.  FWIW, I also specified --enable-threads
when I built it although, again, I have no idea if this works or not.
I've included a ChangeLog of what else has changed in 2.95.3-6 below.
It isn't that much.

All three of these packages were built using my mknetrel scripts.  You
can view the mknetrel configuration files that I used here:

For gcc:
http://sources.redhat.com/cgi-bin/cvsweb.cgi/mknetrel/extra/gcc?rev=1.2&cvsroot=cygwin-apps

For gcc2:
http://sources.redhat.com/cgi-bin/cvsweb.cgi/mknetrel/extra/gcc2?rev=1.1&cvsroot=cygwin-apps

Both of the above reflect the current version of my scripts as of this
writing.  There will undoubtely be newer versions available, so check
the mknetrel repository here:

http://sources.redhat.com/cgi-bin/cvsweb.cgi/mknetrel/?cvsroot=cygwin-apps

for the newest stuff in the "extra" directory.

Before you ask "How did you do X?" in this release, check out the
above mentioned scripts.

So, if you want to play with this, this is what I suggest that you do:

1) Run setup.exe

2) Choose "Install from internet".

3) Get to the "Select Packages" screen.

4) Click on "Exp" on the "Select Packages" screen.

5) Click on the "+" next to the Devel.

6) Click on the arrow thing to the left of the "gcc" until it says 3.1.1-1.

7) Click on the arrow thing to the left of the "gcc-mingw" until it says
   3_1-20020516-1.

8) Click on the arrow thing to the left of the "gcc2" until it says
   2.95.3-6.

Then allow installation to proceed as normal.

This will provide you with a 3.1.1 version of gcc and a 2.95.3-6 version of
gcc-2.

It goes without saying (right?  right?) that you'll have to wait for
these packages to propagate to mirrors before you will be able to
download/install anything.

Please send experiences to the cygwin mailing list, not to me personally.

cgf

gcc2-2.95.3-6 changes not mentioned above:

2002-01-06  Christopher Faylor  <cgf@redhat.com>

        * config/i386/cygwin.h: Fix to allow -nostdinc to drop cygwin includes.
	Only search for cyg library prefix when not -mno-cygwin
	* t-cygwin: Add -lintl as an auxilliary library.
	* x-cygwin: Ditto.

2001-11-05  Danny Smith  <danny_r_smith_2001@yahoo.com.nz>

	* gcov.c (IS_DIR_SEPARATOR): Provide default definition.
	(output_data): Use it to test for absolute pathname.

Fri Oct 29 15:25:07 1999  Arnaud Charlet  <charlet@ACT-Europe.FR>

	* gcov.c (DIR_SEPARATOR): Provide default.
	(output_data): Add test for MS-DOS format absolute filename.
	(fancy_abort): Correct program name.
	(open_files): Open all files in binary mode.
	* libgcc2.c (__bb_exit_func): Likewise.
	* profile.c (init_branch_prob): Specify binary when opening files
--
Please do not send me personal email with cygwin questions.
Use the resources at http://cygwin.com/ .

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Robert Collins]

Just a note: Clicking on 'exp' is not needed to access the new packages,
just clicking on the version number until it shows the desired version is
all thats needed.

Rob


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Cliff Hones]

cgf wrote:

> ...
> Danny Smith has put a lot of work into this release and currently has an
> older version of gcc 3.1 available for testing in mingw.  This 3.1.1
> version incorporates more bug fixes from him as well as from normal gcc
> folk.

Did you really mean that :-).

> gcc 3.1.1 has been built with thread support enabled.  It also includes
> the java compiler but I haven't actually verified that either works.
> Theoretically, this could also include ada support, but if I do decide
> to make this available, it will probably be as a separate package.  I
> may do this with java eventually, too.

Just for the record, Ada should always be capitalised, just like
Cygwin.  [The language is named after Ada Lovelace.]

-- Cliff



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christoph]

> From: Christopher Faylor

...

> Subject: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and
gcc-mingw-3_1-20020516-1


Yeah!


> I've just uploaded a test version of gcc 3.1.1 based on the main gcc
> 3.1.1 branch.
> Danny Smith has put a lot of work into this release and currently has
an
> older version of gcc 3.1 available for testing in mingw.  This 3.1.1
> version incorporates more bug fixes from him as well as from normal
gcc
> folk.
> 
> Danny reports that exception handling should work ok in this release
> although I believe that there are still issues with exception handling
> in dlls.

Hm - being utterly ignorant about Cygwin dll internals, I can at
least report that it seems to be possible to compile (and link
them to the corresponding libstdc++.a and libgcc.a libraries) the
Cygwin dll itself without the standard option -fno-rtti -fno-exceptions.
 

By re-adding the --fno-exceptions  option to the compilation of
winsup/cygwin/lib/_cygwin_crt0_common.cc one seems to also get
a  fully functional  libcygwin.a library - i.e. running the 
test suite resulted in nine failures and no pthread failure. 

(Without re-adding the option -fno-exceptions to the compilation
of _cygwin_crtO_common.cc one gets a link error ala
.. libcygwin.a(_cygwin_crt0_common.o):
In function `_cygwin_crt0_common':
.. /src/winsup/cygwin/lib/_cygwin_crt0_common.cc:41: 
undefined reference to `___gxx_personality_v0'.)

> 
...
> gcc 3.1.1 has been built with thread support enabled.  It also
includes
> the java compiler but I haven't actually verified that either works.
> Theoretically, this could also include ada support, but if I do decide

Cool ... using basically the same options and sources I was able to 
compile the Ada extension + libs + tools (the compilation of gnatdll
required a minor tweak).
 
> to make this available, it will probably be as a separate package.  I
> may do this with java eventually, too.

I take that that it is still not possible to compile libgjc?

> 
...

> 2002-01-06  Christopher Faylor  <cgf@redhat.com>
> 
>         * config/i386/cygwin.h: Fix to allow -nostdinc to drop cygwin
includes.
> 	Only search for cyg library prefix when not -mno-cygwin
> 	* t-cygwin: Add -lintl as an auxilliary library.
> 	* x-cygwin: Ditto.

For my native bootstrap I had to explicitly copy cygming.h into
the build directory  Objdir/gcc/gcc to make things work.


/Christoph 


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Nicholas Wourms]

--- Christopher Faylor <cgf-noreply@cygwin.com> wrote:
[SNIP]
> My main contributions are some configury reorg and of course, my old
> friend, "-mno-cygwin".  In theory, "-mno-cygwin" should work much
> better in this release.  It should even be functional for C++.
> To that end, I'm also releasing an auxiliary package called
> gcc-mingw-3_1-20020516-1.  This package contains header and library
> files from the mingw release from the main mingw site, repackaged with
> only the bits that are required for the cygwin version of gcc.  I
> hope to be updating this as necessary (or getting someone to volunteer
> to do this once things have settled down).

Do we still need to create a "g++-3" symlink in the /usr/include/mingw dir
that points to /usr/include/g++-3/ for c++ to work?  Or is that covered by
the gcc-mingw-3_1-20020516-1 package?

[SNIP]
> That brings us to gcc2-2.95.3-6.  As you might guess, this package is
> just a repackaging of the old gcc-2.95.3 release.  The difference here,
> though, is that all of the appropriate files are labelled gcc-2.exe
> g++-2.exe gcc-2.man, etc.  The reason for this is to allow you to use
> both the old and new versions of gcc on your system at the same time.
> Then, when you're satisfied that you no longer require gcc2 stuff you
> can uninstall the gcc2 package without affecting the gcc 3.1.1 stuff.
> Or, you can just keep both around.

Will it be as simple as passing CC=gcc-2 and CXX=g++-2 to the configure
scripts?

[SNIP]
> All three of these packages were built using my mknetrel scripts.  You
> can view the mknetrel configuration files that I used here:
> 
> For gcc:
>
http://sources.redhat.com/cgi-bin/cvsweb.cgi/mknetrel/extra/gcc?rev=1.2&cvsroot=cygwin-apps

Small nitpick, but according to the gcc install.texi, the proper
enable-languages flag for fortran support is "f77" not "f".
 
> For gcc2:
>
http://sources.redhat.com/cgi-bin/cvsweb.cgi/mknetrel/extra/gcc2?rev=1.1&cvsroot=cygwin-apps
[SNIP]

Again, thanks for the hard work and diligence.  I do have a few minor
points to contend with.

1)Objective-C or the lack thereof.  Both David-B and Wayne have reported
that they have had success compiling the Objective-C compiler and runtime
libraries.  There has been reports from others as well, most recently in:

http://sources.redhat.com/ml/cygwin/2002-07/msg00229.html
 
Would it be possible to add "objc" to the "--enable-languages="?  Since it
seems to compile cleanly OOB, I think it should be added so that those
wishing to develop in objc on cygwin may do so.

2)Since you wanted to enable java support, you must pass a few more flags
for it to work.  First off, the java runtime library is disabled by
default on most every platform.  Also, as per David B's reccomendation,
you should explicitly pass the sjlj-exceptions flag during configuration. 
So, the most optimal set of flags to pass would be:

--enable-sjlj-exceptions --enable-libgcj --enable-interpreter

If you run into errors in the libjava dir, you might want to check its
config.h file for proper header/library detection.

Other than that, congratulations on a job well done and thanks!

Cheers,
Nicholas

P.S. - That sedit line at the end of the mknetrel script had me rolling on
the floor!  ;-)

__________________________________________________
Do You Yahoo!?
Sign up for SBC Yahoo! Dial - First Month Free
http://sbc.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Gunnar Degnbol]

Christopher Faylor wrote:


>> I've just uploaded a test version of gcc 3.1.1 based on the main gcc
3.1.1 branch.

I tried the new __fastcall support, but the assembler choked on symbols
beginning with '@':

compiling (gcc -c -save-temps fasttest.c) this file:

void __fastcall FastFunc(int parm1)
{
}

void (__fastcall * func)(int parm1) = FastFunc;

produces this assembly output:

	.file	"fasttest.c"
	.text
	.align 2
.globl @FastFunc@4
	.def	@FastFunc@4;	.scl	2;	.type	32;	.endef
@FastFunc@4:
	pushl	%ebp
	movl	%esp, %ebp
	subl	$4, %esp
	movl	%ecx, -4(%ebp)
	leave
	ret
.globl _func
	.data
	.align 4
_func:
	.long	@FastFunc@4

which gives these errors:

fasttest.s: Assembler messages:
fasttest.s:4: Warning: rest of line ignored; first ignored character is `@'
fasttest.s:5: Warning: rest of line ignored; first ignored character is `@'
fasttest.s:6: Warning: rest of line ignored; first ignored character is `@'
fasttest.s:17: Error: bad expression
fasttest.s:17: Warning: rest of line ignored; first ignored character is `F'


The file compiles without problems in VC++.


Gunnar



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sun, Jul 07, 2002 at 09:46:46AM +0100, Cliff Hones wrote:
>> Danny Smith has put a lot of work into this release and currently has an
>> older version of gcc 3.1 available for testing in mingw.  This 3.1.1
>> version incorporates more bug fixes from him as well as from normal gcc
>> folk.
>
>Did you really mean that :-).

And, did you really mean to send specious comments to thousands of people?

>> gcc 3.1.1 has been built with thread support enabled.  It also includes
>> the java compiler but I haven't actually verified that either works.
>> Theoretically, this could also include ada support, but if I do decide
>> to make this available, it will probably be as a separate package.  I
>> may do this with java eventually, too.
>
>Just for the record, Ada should always be capitalised, just like
>Cygwin.  [The language is named after Ada Lovelace.]

Wow.  This is vitally important.  It's crucial that everyone be apprised
of this crucial bit of knowledge.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sun, Jul 07, 2002 at 01:30:58PM +0200, Christoph wrote:
>> From: Christopher Faylor
>> although I believe that there are still issues with exception handling
>> in dlls.
>
>Hm - being utterly ignorant about Cygwin dll internals, I can at
>least report that it seems to be possible to compile (and link
>them to the corresponding libstdc++.a and libgcc.a libraries) the
>Cygwin dll itself without the standard option -fno-rtti -fno-exceptions.

Huh?  It's always been possible to do that.  The options are included
only avoid any overhead from including them in an application
(cygwin1.dll) that didn't need or use them.

>> to make this available, it will probably be as a separate package.  I
>> may do this with java eventually, too.
>
>I take that that it is still not possible to compile libgjc?

I haven't tried.

>> 2002-01-06  Christopher Faylor  <cgf@redhat.com>
>> 
>>         * config/i386/cygwin.h: Fix to allow -nostdinc to drop cygwin
>includes.
>> 	Only search for cyg library prefix when not -mno-cygwin
>> 	* t-cygwin: Add -lintl as an auxilliary library.
>> 	* x-cygwin: Ditto.
>
>For my native bootstrap I had to explicitly copy cygming.h into
>the build directory  Objdir/gcc/gcc to make things work.

I don't see how that could possibly be necessary.  The only reference
to cygming.h should b as i386/cygming.h.  Copying it to your build directory
would have no effect.  It should only be referenced in the config.h file
as #include "i386/cygming.h".

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sun, Jul 07, 2002 at 05:23:42AM -0700, Nicholas Wourms wrote:
>Do we still need to create a "g++-3" symlink in the /usr/include/mingw
>dir that points to /usr/include/g++-3/ for c++ to work?  Or is that
>covered by the gcc-mingw-3_1-20020516-1 package?

That would be a question that would easily be satisfied by empirical
testing, wouldn't it?

>Would it be possible to add "objc" to the "--enable-languages="?  Since it
>seems to compile cleanly OOB, I think it should be added so that those
>wishing to develop in objc on cygwin may do so.

Nope.  I am not interested in supporting objective c at this point.  If I
was it would be a separate package anyway.

>2)Since you wanted to enable java support, you must pass a few more flags
>for it to work.  First off, the java runtime library is disabled by
>default on most every platform.  Also, as per David B's reccomendation,
>you should explicitly pass the sjlj-exceptions flag during configuration. 
>So, the most optimal set of flags to pass would be:
>
>--enable-sjlj-exceptions --enable-libgcj --enable-interpreter

--enable-sjlj-exceptions is not supposed to be required anymore.  I'll
try turning on --enable-libgcj.  Meant to do that previously.  Getting this
working were the impetus for David Billinghurst's changes.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Nicholas Wourms]

--- Christopher Faylor <cgf@redhat.com> wrote:
> 
> Nope.  I am not interested in supporting objective c at this point.  If
> I was it would be a separate package anyway.

Well will there be a point at which you will be?  Porting some, if not
all, of the GNUStep environment can't happen until you do so.  Hopefully
when you see it fit to provide Ada support, you will provide the
objective-c compiler.  If you don't want to support it ever, will you
allow someone else to package it and submit for "official" inclusion?

> --enable-sjlj-exceptions is not supposed to be required anymore.  I'll
> try turning on --enable-libgcj.  Meant to do that previously.  Getting
> this
> working were the impetus for David Billinghurst's changes.

Thanks, I appreciate that!

There being no commentary from you on the f77 issue, I assume you caught
that.

Cheers,
Nicholas

__________________________________________________
Do You Yahoo!?
Sign up for SBC Yahoo! Dial - First Month Free
http://sbc.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Charles Wilson]

[-- Attachment #1: Type: text/plain, Size: 863 bytes --]

I believe there are a few problems with gcc-2's c++ support -- I've 
attached an example that should demonstrate the problem.  (just unpack, 
cd, and make).

I'm getting lots of these sorts of errors:

/tmp/dllhelpers-0.4.0/cxx/usedll.cc:22: undefined reference to 
`endl(ostream &)'

I think the problem is that g++-2 is using g++(3.1.x)'s libstdc++.a 
library.  running the link command with -verbose, and I get:

...
attempt to open /usr/local/lib/libstdc++.a failed
attempt to open /usr/lib/libstdc++.dll.a failed
attempt to open /usr/lib/stdc++.dll.a failed
attempt to open /usr/lib/libstdc++.a succeeded
attempt to open .//libgcc.dll.a failed
attempt to open .//gcc.dll.a failed
...

But, didn't the C++ ABI change between gcc 2.95.x and 3.x ?  If so, does 
this mean that g++(3.x) can't use any C++ libraries that were compiled 
with g++(2.95.x)?

--Chuck

[-- Attachment #2: cxx.tar.bz2 --]
[-- Type: application/octet-stream, Size: 1927 bytes --]

[-- Attachment #3: Type: text/plain, Size: 214 bytes --]

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sun, Jul 07, 2002 at 10:46:14PM -0400, Charles Wilson wrote:
>I believe there are a few problems with gcc-2's c++ support -- I've 
>attached an example that should demonstrate the problem.  (just unpack, 
>cd, and make).
>
>I'm getting lots of these sorts of errors:
>
>/tmp/dllhelpers-0.4.0/cxx/usedll.cc:22: undefined reference to 
>`endl(ostream &)'

Thanks for the test case.  I see these too.  I know what's going on.
I should have a spec file patch for this for you to try.

I have to admit that the only c++ programming that I do is for cygwin,
which is obviously a very special case, since building cygwin and its
utilities worked fine.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Charles Wilson]

Christopher Faylor wrote:


> 
> Thanks for the test case.  I see these too.  I know what's going on.


I'm glad you do -- because my guess was wrong.  I see the same errors 
using regular g++ (3.1.1).


> I should have a spec file patch for this for you to try.


'Kay.


> I have to admit that the only c++ programming that I do is for cygwin,
> which is obviously a very special case, since building cygwin and its
> utilities worked fine.


<g>  Ain't that always the way?

This is why we have 'test releases'.  :-)

--Chuck



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Mon, Jul 08, 2002 at 12:15:01AM -0400, Charles Wilson wrote:
>
>
>Christopher Faylor wrote:
>
>
>>
>>Thanks for the test case.  I see these too.  I know what's going on.
>
>
>I'm glad you do -- because my guess was wrong.  I see the same errors 
>using regular g++ (3.1.1).

I don't see exactly the same errors.  I can't even compile your stuff
with g++.  :-( It appears to be due to --enable-version-specific-runtime-libs,
AFAICT.

I'll respin both packages ASAP.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Charles Wilson]

Christopher Faylor wrote:

> I don't see exactly the same errors.  I can't even compile your stuff
> with g++.  :-( It appears to be due to --enable-version-specific-runtime-libs,
> AFAICT.


Right -- I had to change #include <iostream.h> to <iostream>, etc.  I'm 
not real familiar with the source-level changes that gcc-3.x *forces* 
onto people -- but I'm not real happy about them.  The price of 
progress, I suppose.

Does anybody know of a good reference, along the lines of "What changed 
in g++ between gcc-2.95.3 and gcc-3.x"?  I'm not talking about the 
innards of gcc; but rather, user-visible changes that force me to change 
my perfectly good 2.95.3 C++ code so that it'll compile with g++ 3.x.

--Chuck


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Nicholas Wourms]

--- Charles Wilson <cwilson@ece.gatech.edu> wrote:
> 
> 
> Christopher Faylor wrote:
> 
> > I don't see exactly the same errors.  I can't even compile your stuff
> > with g++.  :-( It appears to be due to
> --enable-version-specific-runtime-libs,
> > AFAICT.
> 
> 
> Right -- I had to change #include <iostream.h> to <iostream>, etc.  I'm 
> not real familiar with the source-level changes that gcc-3.x *forces* 
> onto people -- but I'm not real happy about them.  The price of 
> progress, I suppose.
> 
> Does anybody know of a good reference, along the lines of "What changed 
> in g++ between gcc-2.95.3 and gcc-3.x"?  I'm not talking about the 
> innards of gcc; but rather, user-visible changes that force me to change
> 
> my perfectly good 2.95.3 C++ code so that it'll compile with g++ 3.x.
> 
> --Chuck

The changes are not to gcc, but to libstdc++.  #include <foo.h>, from what
I gather, is a non-standard relic of libstdc++ v2.  Apparently in v3, they
discourage this behavior.. (but it should just spit out a warning, not
fail to compile).  FWIW, you might want to check out:

http://gcc.gnu.org/onlinedocs/libstdc++/17_intro/howto.html
^---This is where your <foo> vs. <foo.h> is discussed
http://gcc.gnu.org/onlinedocs/libstdc++/faq/

Not being a c++ person myself, I found a lot of useful information there
that I didn't know.  Hope that helps!

Cheers,
Nicholas

__________________________________________________
Do You Yahoo!?
Sign up for SBC Yahoo! Dial - First Month Free
http://sbc.yahoo.com

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Charles Wilson]

Chris --

I know that it is NOT yet time for ObjC or Ada or Java or any of those 
"extra" front ends (especially since g++ isn't even working yet <g>). 
However (you *knew* there was a however), gcc-2.95.3-5 contained g77 -- 
but gcc2-2.95.3-6 doesn't.

It probably should contain it, tho, just to minimize disruption when 
gcc2/gcc-3.1.x goes "gold".

--Chuck


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Shawn Behrens]

Hi,

>My main contributions are some configury reorg and of course, my old
>friend, "-mno-cygwin".  In theory, "-mno-cygwin" should work much
>better in this release.  It should even be functional for C++.

Okay, I'm knuckle-headed. Consider this example:

gcc-2 -mno-cygwin hello.c

No trouble, an a.exe containing a Hello, World!

gcc -mno-cygwin hello.c

gcc: installation problem, cannot exec `cc1': No such file or directory

Alright, then, add /usr/lib/gcc-lib/i686-pc-cygwin/3.1.1/ to the end of the 
$PATH

ld: cannot open crt2.o: No such file or directory

At which point I am stumped. What am I, knucklehead that I am, missing?

Shawn



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Mon, Jul 08, 2002 at 11:13:36PM -0400, Shawn Behrens wrote:
>>My main contributions are some configury reorg and of course, my old
>>friend, "-mno-cygwin".  In theory, "-mno-cygwin" should work much
>>better in this release.  It should even be functional for C++.
>
>Okay, I'm knuckle-headed. Consider this example:
>
>gcc-2 -mno-cygwin hello.c
>
>No trouble, an a.exe containing a Hello, World!
>
>gcc -mno-cygwin hello.c
>
>gcc: installation problem, cannot exec `cc1': No such file or directory
>
>Alright, then, add /usr/lib/gcc-lib/i686-pc-cygwin/3.1.1/ to the end of the 
>$PATH
>
>ld: cannot open crt2.o: No such file or directory
>
>At which point I am stumped. What am I, knucklehead that I am, missing?

Did you install the cygwin-mingw package?  It doesn't sound like you did.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Shawn Behrens]

>Did you install the cygwin-mingw package?  It doesn't sound like you did.

Umm ... duh. Thank-you.

Shawn



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Charles Wilson]

Paul Butcher wrote:

> 
> Can I put another vote in for this one - 


No.  :-)

> the lack of objC in recent cygwin 
> gcc builds has been a real problem forcing people to use the mingw gcc.  I've 
> recently compiled gcc V3.1 with cygwin including objC with no problem so 
> there doesn't seem to be much "support" needed.  The size added to the 
> overall package is negligable, certainly compared to F77 which probably ought 
> to be separately packaged, like ADA.
> 
> (It works even better if you replace the libobjc with that from GNUstep as 
> this allows DLLs and makes the runtime a DLL.)
> 
> Otherwise well done for getting gcc 3.x into cygwin build at last - look 
> forward to the fully released version when ready.  Will be interesting if you 
> can get libgcj working too as you're currently trying to do.


It's too early to worry about ADDITIONAL front ends.  Let's make sure 
that the CURRENT front ends provided by gcc-2.95.3-5 will not be BROKEN 
  by an upgrade to the cygwin-special gcc-3.1.1.

That means gcc, g++ (maybe fortran), since ONLY those three front ends 
are provided by cygwin-special gcc-2.95.3-5.

After the bugs in THOSE frontends are squashed, then we can *talk* about 
additional frontends, but not until then.

(IMO, the current focus on libgcj/java frontend is misguided, for the 
same reasons...but hey, that's just my opinion)

--Chuck




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Paul Butcher]

Nicholas Wourms <nwourms at yahoo dot com> wrote:
>--- Christopher Faylor <cgf@redhat.com> wrote:
>> 
>> Nope.  I am not interested in supporting objective c at this point.  If
>> I was it would be a separate package anyway.
>
> Well will there be a point at which you will be?  Porting some, if not
> all, of the GNUStep environment can't happen until you do so.  Hopefully
> when you see it fit to provide Ada support, you will provide the
> objective-c compiler.  If you don't want to support it ever, will you
> allow someone else to package it and submit for "official" inclusion?

Can I put another vote in for this one - the lack of objC in recent cygwin 
gcc builds has been a real problem forcing people to use the mingw gcc.  I've 
recently compiled gcc V3.1 with cygwin including objC with no problem so 
there doesn't seem to be much "support" needed.  The size added to the 
overall package is negligable, certainly compared to F77 which probably ought 
to be separately packaged, like ADA.

(It works even better if you replace the libobjc with that from GNUstep as 
this allows DLLs and makes the runtime a DLL.)

Otherwise well done for getting gcc 3.x into cygwin build at last - look 
forward to the fully released version when ready.  Will be interesting if you 
can get libgcj working too as you're currently trying to do.

Paul Butcher
Alton, Hants, UK

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dockeen]

I love Ada as much as any man....oops...what I meant was I can continue to
use
GNAT as a stand-alone tool.  (I'm sure that make you folks feel SO much
better)

g77 on the other hand....

Ack, I'm ASKING for FORTRAN.....I have lost it.

Everyone, laugh at the silly physicist!

Wayne Keen


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dockeen]

Charles,

I know when I have used my own home cooked version of 3.1 and 3.1.1, 
when I compile, I have to "push" it down the correct library path, i.e.

/mygcc/bin/g++ -o hello hello.cpp -L/mygcc/lib

or I will get some errors that I had attributed to some kind of cross-
library foul-up.

General,

I have not seen any traffic on the g77 issue (appears to not be built),
just wanted to make sure someone caught what I was mumbling about
earlier today...

Take care folks! 

Wayne Keen

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dylan Cuthbert]

Has anyone tried gdb 5.2 (or 5.1.1) with any code generated by these new
versions?  Good results?  I'm going to migrate to these new test versions on
Thursday or Friday but I'd like to know beforehand if anyone has had any
problems with gdb.

Regards

--
---------------------------------
Q-Games, Dylan Cuthbert.
http://www.q-games.com

"Dockeen" <dockeen@mchsi.com> wrote in message
news:NCBBLKKJGBJOLDFMCKFPKEKOCAAA.dockeen@mchsi.com...
> Probably something the knucklehead is doing.  Setup "sees" the gcc-2
> and mingw, but it won't toggle to 3.1.1-1.  This is on the rcn sites,
> and I did check, the files are there, but all the gcc toggle ever shows
> is options for uninstalling / reinstalling / skip for 2.95.3-5 version.
>
> Probably just brain damage on my part...
>
> Wayne Keen
>
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
>
>




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dockeen]

That site worked like a champ, gcc -v gives (among other things)

gcc version 3.1.1 20020706 (prerelease)

Major coolies, I've been rolling my own for a while, thanks!

I'll play with some code in the morning.

Wayne

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sat, Jul 06, 2002 at 11:31:50PM -0400, Dockeen wrote:
>Criminal stupidity on my part.  Thanks for more consideration that I
>deserve Chris.

Give yourself a break.  It wasn't obvious.  I played with the little
button in setup.exe for a while I should before I figured out what
was wrong.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sun, Jul 07, 2002 at 12:25:58AM -0400, Christopher Faylor wrote:
>On Sat, Jul 06, 2002 at 11:14:43PM -0400, Dockeen wrote:
>>Probably something the knucklehead is doing.  Setup "sees" the gcc-2
>>and mingw, but it won't toggle to 3.1.1-1.  This is on the rcn sites,
>>and I did check, the files are there, but all the gcc toggle ever shows
>>is options for uninstalling / reinstalling / skip for 2.95.3-5 version.
>>
>>Probably just brain damage on my part...
>
>mirrors.rcn.net has the .tar.bz2 files but not the updated setup.ini.
>If you look at the setup.ini on that site, you'll see that it doesn't
>list gcc 3.1.1.  The setup.ini on sources.redhat.com definitely does
>have 3.1.1.
>
>So the only solution is to wait...

I just checked the ftp transfer logs and it appears that ftp.nas.nasa.gov
has the up-to-date files.  It's pretty slow, though.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dockeen]

Criminal stupidity on my part.
Thanks for more consideration that I deserve Chris.

Waynebozo

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Christopher Faylor]

On Sat, Jul 06, 2002 at 11:14:43PM -0400, Dockeen wrote:
>Probably something the knucklehead is doing.  Setup "sees" the gcc-2
>and mingw, but it won't toggle to 3.1.1-1.  This is on the rcn sites,
>and I did check, the files are there, but all the gcc toggle ever shows
>is options for uninstalling / reinstalling / skip for 2.95.3-5 version.
>
>Probably just brain damage on my part...

mirrors.rcn.net has the .tar.bz2 files but not the updated setup.ini.
If you look at the setup.ini on that site, you'll see that it doesn't
list gcc 3.1.1.  The setup.ini on sources.redhat.com definitely does
have 3.1.1.

So the only solution is to wait...

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Available for test: gcc-3.1.1-1, gcc2-2.95.3-6, and gcc-mingw-3_1-20020516-1

[Dockeen]

Probably something the knucklehead is doing.  Setup "sees" the gcc-2
and mingw, but it won't toggle to 3.1.1-1.  This is on the rcn sites,
and I did check, the files are there, but all the gcc toggle ever shows
is options for uninstalling / reinstalling / skip for 2.95.3-5 version.

Probably just brain damage on my part...

Wayne Keen

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/