Added branch for compiling with c++-compat

Added branch for compiling with c++-compat

[Matt Rice]

Hi all, I added a branch for compiling gdb with c++-compat
named: archer-ratmice-compile-Wc++-compat

I've been doing:
CFLAGS="-Wc++-compat" ./configure --disable-werror
and then tackling things one type of warning at a time.
in hopes that this will make the patches more easily reviewable.

The downside to this I suppose is that I may temporarily introduce a
warning and then
subsequently miss it in the deluge... If this happens i'll fix these
as I notice.

I guess a balance between number of patches and patch length will be
somewhat important....
I've just been commiting about at the point that I get the urge to
hold down the space bar
and see how much longer the dang patch is.

so far everything has been fairly mechanical, there is one ada macro
which looks like it may be remotely interesting.

Re: Added branch for compiling with c++-compat

[Matt Rice]

On Fri, Jun 11, 2010 at 6:52 PM, Matt Rice <ratmice@gmail.com> wrote:
> Hi all, I added a branch for compiling gdb with c++-compat
> named: archer-ratmice-compile-Wc++-compat

so this branch now compiles with -Wc++-compat with -Werror enabled,
and has caught up to gdb/master

due to some weirdness in the master branch you've got to
--disable-gdbtk when running configure.
and it requires a patch i've sent to binutils

http://sourceware.org/ml/binutils/2010-09/msg00265.html

i'm going to start doing testing on various platforms available on the
gcc compile farm, then start preparing patches to the list.

Re: Added branch for compiling with c++-compat

[Keith Seitz]

On 09/16/2010 02:18 PM, Matt Rice wrote:
> due to some weirdness in the master branch you've got to
> --disable-gdbtk when running configure.

What's happening? Does this occur on master, too?

Keith

Re: Added branch for compiling with c++-compat

[Matt Rice]

On Thu, Sep 16, 2010 at 2:38 PM, Keith Seitz <keiths@redhat.com> wrote:
> On 09/16/2010 02:18 PM, Matt Rice wrote:
>>
>> due to some weirdness in the master branch you've got to
>> --disable-gdbtk when running configure.
>
> What's happening? Does this occur on master, too?
>
> Keith

I actually haven't tried, but I imagine so,
'make clean' fails to clean in the gdb/testsuite/gdb.gdbtk
directory it looks like the directory appeared after a recent commit

It may only happen after running make check.

Re: Added branch for compiling with c++-compat

[Keith Seitz]

On 09/16/2010 02:43 PM, Matt Rice wrote:
> I actually haven't tried, but I imagine so,
> 'make clean' fails to clean in the gdb/testsuite/gdb.gdbtk
> directory it looks like the directory appeared after a recent commit
>
> It may only happen after running make check.

I've updated master and tried this. I am not seeing any errors. Tried 
make check, too. gdb/testsuite/gdb.gdbtk does not appear on master, 
either (although it is still in CVS).

If you can find the right formula for reproducing this, let me know.

Keith

Re: Added branch for compiling with c++-compat

[Matt Rice]

On Thu, Sep 16, 2010 at 3:49 PM, Keith Seitz <keiths@redhat.com> wrote:
> On 09/16/2010 02:43 PM, Matt Rice wrote:
>>
>> I actually haven't tried, but I imagine so,
>> 'make clean' fails to clean in the gdb/testsuite/gdb.gdbtk
>> directory it looks like the directory appeared after a recent commit
>>
>> It may only happen after running make check.
>
> I've updated master and tried this. I am not seeing any errors. Tried make
> check, too. gdb/testsuite/gdb.gdbtk does not appear on master, either
> (although it is still in CVS).
>
> If you can find the right formula for reproducing this, let
me know.

thats weird that you're not seeing it, you are using git or cvs?

http://sourceware.org/git/?p=archer.git;a=tree;f=gdb/testsuite/gdb.gdbtk;h=ebbdaf8af5b184f6a4a1ed7ee387c7b199e5f407;hb=cf7b7b38fd5f32865f0c91d1b6f929c08e04ff0a

http://sourceware.org/git/?p=gdb.git;a=tree;f=gdb/testsuite/gdb.gdbtk;h=ebbdaf8af5b184f6a4a1ed7ee387c7b199e5f407;hb=cf7b7b38fd5f32865f0c91d1b6f929c08e04ff0a

http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/testsuite/gdb.gdbtk/?cvsroot=src

make[5]: Entering directory `/home/ratmice/git/gdb/gdb/testsuite/gdb.gdbtk'
make[5]: *** No rule to make target `clean'.  Stop.
make[5]: Leaving directory `/home/ratmice/git/gdb/gdb/testsuite/gdb.gdbtk'
make[4]: *** [clean] Error 2

[ratmice@willow gdb]$ git diff
[ratmice@willow gdb]$ git branch
* master

all i did to reproduce was

 ./configure --disable-gdbtk && make distclean && ./configure && make
&& make clean

[ratmice@willow gdb]$ git checkout 3307a59dab65689ff1cad93f78a3f0ef34b7ddb0

[ratmice@willow gdb]$ ls gdb/testsuite/gdb.gdbtk
ls: cannot access gdb/testsuite/gdb.gdbtk: No such file or directory

[ratmice@willow gdb]$ git checkout 76d14d8e5fbdcd8381b9b89042f2a3b1e433e431

[ratmice@willow gdb]$ ls gdb/testsuite/gdb.gdbtk/
browser.exp  console.exp       c_variable.exp  windows.exp
ChangeLog    cpp_variable.exp  srcwin.exp

so the 2010-08-10 commit managed to pull in all the history from the
files the commit modified, the rest of the gdb.gdbtk files which were
unmodified by that commit don't appear to be in the git repositories.
(and Makefile.in is not one of them so make clean fails).

i imagine someone with commit access to the gdb git repository just
needs to delete the gdb/testsuite/gdb.gdbtk directory.