Configure GDB for MinGW

Configure GDB for MinGW

[Manu]

[-- Attachment #1: Type: text/plain, Size: 166 bytes --]

The attached patch permits to configure GDB under MinGW.

Manu.
Attachment:
config-05.diff
Description: Text document
Attachment:
ChangeLog
Description: Binary data


[-- Attachment #2: ChangeLog --]
[-- Type: text/plain, Size: 445 bytes --]

2004-05-21  Manu B  <manubee@users.sourceforge.net>

	* Makefile.in: Build ser-mingw32.c under MinGW.
	* config/i386/mingw32.mh: New file.
	* config/i386/mingw32.mt: Ditto.
	* config/i386/nm-mingw32.h: Ditto.
	* config/i386/tm-mingw32.h: Ditto.
	* config/i386/xm-mingw32.h: Ditto.
	* configure.host: Add MinGW host.
	* configure.in: Disable gdbtk under MinGW, define SER_HARDWIRE.
	* configure.tgt: Add MinGW target.
	* ser-mingw32.c: New file.

Re: Configure GDB for MinGW

[Mark Kettenis]

   From: "Manu" <manubee@wanadoo.fr>
   Date: Fri, 21 May 2004 12:04:20 +0200

Thanks for your contribution.  I think it is small enough to be
accepted without a copyright assignment, especially if you follow the
advice below.  I'm not completely sure though.

A few remarks and questions:

1. Please update the copyright year of new files or files that you changed.

2. Please remove any empty variables in your .mh and .mt files.

3. Is MH_CFLAGS=-D_POSIX in mingw32.mh really necessary?

4. Your nm-mingw32.h file seems to be a verbatim copy if nm-cygwin.h.
   Please use nm-cygwin.h instead.

5. All your tm-mingw32.h file does in including tm-cygwin.h.  Actually
   I think you can just use cygwin.mt instead of mingw32.mt.  I expect
   that MinGW uses the same convention for returning structs as
   Cygwin, so that will actually fix a few bugs.

6. Why is the line

   #define getkey getch

   in xm-mingw32.h needed?

7. Looks like ser-mingw32.c is doing nothing.  AFAICT it shouldn't be
   necessary.

Cheers,

Mark

Re: Configure GDB for MinGW

[Christopher Faylor]

On Mon, May 24, 2004 at 07:36:51PM +0200, Mark Kettenis wrote:
>   From: "Manu" <manubee@wanadoo.fr>
>   Date: Fri, 21 May 2004 12:04:20 +0200
>
>Thanks for your contribution.  I think it is small enough to be
>accepted without a copyright assignment, especially if you follow the
>advice below.  I'm not completely sure though.

I guess we'd have to see the actual finished patch but given that it makes
changes to a number of files and adds new functionality, I don't think
it really could be considered trivial.

>A few remarks and questions:
>
>1. Please update the copyright year of new files or files that you changed.
>
>2. Please remove any empty variables in your .mh and .mt files.
>
>3. Is MH_CFLAGS=-D_POSIX in mingw32.mh really necessary?
>
>4. Your nm-mingw32.h file seems to be a verbatim copy if nm-cygwin.h.
>   Please use nm-cygwin.h instead.

For consistency with gcc, we could rename this file n-cygming.h.

>5. All your tm-mingw32.h file does in including tm-cygwin.h.  Actually
>   I think you can just use cygwin.mt instead of mingw32.mt.  I expect
>   that MinGW uses the same convention for returning structs as
>   Cygwin, so that will actually fix a few bugs.

Again, the cygwin.mt file could be renamed.

cgf

Re: Configure GDB for MinGW

[Manu]

Christopher Faylor wrote:

> On Mon, May 24, 2004 at 07:36:51PM +0200, Mark Kettenis wrote:
> >   From: "Manu"
> >   Date: Fri, 21 May 2004 12:04:20 +0200
> >
> >Thanks for your contribution.  I think it is small enough to be
> >accepted without a copyright assignment, especially if you follow the
> >advice below.  I'm not completely sure though.
>
> I guess we'd have to see the actual finished patch but given that it makes
> changes to a number of files and adds new functionality, I don't think
> it really could be considered trivial.

Is it true that the FSF sometimes accepts a disclaimer instead of a
copyright
assignment?

I've been managing a project for two or three years, I created the
application
under the GNU GPL and its library under the public domain.
I'm supporting both Free Software and Open Source. Then, I think you can
trust what I'm providing. IMO, It is more valuable than any paper.

> >A few remarks and questions:
> >
> >1. Please update the copyright year of new files or files that you
changed.
> >
> >2. Please remove any empty variables in your .mh and .mt files.
> >
> >3. Is MH_CFLAGS=-D_POSIX in mingw32.mh really necessary?
> >
> >4. Your nm-mingw32.h file seems to be a verbatim copy if nm-cygwin.h.
> >   Please use nm-cygwin.h instead.
>
> For consistency with gcc, we could rename this file n-cygming.h.
>
> >5. All your tm-mingw32.h file does in including tm-cygwin.h.  Actually
> >   I think you can just use cygwin.mt instead of mingw32.mt.  I expect
> >   that MinGW uses the same convention for returning structs as
> >   Cygwin, so that will actually fix a few bugs.
>
> Again, the cygwin.mt file could be renamed.

I like the idea.

Manu.

Re: Configure GDB for MinGW

[Christopher Faylor]

On Tue, May 25, 2004 at 11:43:42PM +0200, Manu wrote:
>Christopher Faylor wrote:
>
>> On Mon, May 24, 2004 at 07:36:51PM +0200, Mark Kettenis wrote:
>> >   From: "Manu"
>> >   Date: Fri, 21 May 2004 12:04:20 +0200
>> >
>> >Thanks for your contribution.  I think it is small enough to be
>> >accepted without a copyright assignment, especially if you follow the
>> >advice below.  I'm not completely sure though.
>>
>> I guess we'd have to see the actual finished patch but given that it makes
>> changes to a number of files and adds new functionality, I don't think
>> it really could be considered trivial.
>
>Is it true that the FSF sometimes accepts a disclaimer instead of a
>copyright assignment?

I believe they do sometimes.

>I've been managing a project for two or three years, I created the
>application under the GNU GPL and its library under the public domain.
>I'm supporting both Free Software and Open Source.  Then, I think you
>can trust what I'm providing.  IMO, It is more valuable than any paper.

Judging your reputation isn't really an option for me.  I work for TimeSys.
If you want to take up the matter with the FSF and let me know what they
think I can wait for that.  The paperwork isn't that odious and signing
it would mean that you can make future hassle-free contributions to gdb.

In the meantime, I'll wait on your patch.  Maybe it really could be
considered trivial.  We could always get a ruling from Andrew.  It still
seems like the best long term solution would be to get the assignment
in with the FSF, though.

I'm not quite sure how you could do the cygming stuff that I mentioned
in a patch but if you just supply patches against cygwin named files,
and use cygming in the configuration, I'll figure things out somehow.

cgf

Re: Configure GDB for MinGW

[Eli Zaretskii]

> Date: Tue, 25 May 2004 22:27:49 -0400
> From: Christopher Faylor <cgf@alum.bu.edu>
> 
> >I've been managing a project for two or three years, I created the
> >application under the GNU GPL and its library under the public domain.
> >I'm supporting both Free Software and Open Source.  Then, I think you
> >can trust what I'm providing.  IMO, It is more valuable than any paper.
> 
> Judging your reputation isn't really an option for me.

Signing papers is not about the reputation of the person who
contributes; in fact, it has nothing to do with that at all.  It is
required for the GNU project to be able to defend in court its legal
right to distribute every single line of the code in its projects, in
the (arguably improbable) case that some software vendor will try to
claim that part of the GDB code was unlawfully stolen from that
software vendor.

For example, your employer could claim that code you contributed to
GDB was written when you were working for that employer and that
therefore that code is your employer's property, and the FSF has no
right to distribute it.

Re: Configure GDB for MinGW

[Manu]

Mark Kettenis wrote:


> Thanks for your contribution.  I think it is small enough to be
> accepted without a copyright assignment, especially if you follow the
> advice below.  I'm not completely sure though.
>
> A few remarks and questions:
>
> 1. Please update the copyright year of new files or files that you
changed.
>
> 2. Please remove any empty variables in your .mh and .mt files.

Ok.

> 3. Is MH_CFLAGS=-D_POSIX in mingw32.mh really necessary?

I need to check that point.

> 4. Your nm-mingw32.h file seems to be a verbatim copy if nm-cygwin.h.
>    Please use nm-cygwin.h instead.
>
> 5. All your tm-mingw32.h file does in including tm-cygwin.h.  Actually
>    I think you can just use cygwin.mt instead of mingw32.mt.  I expect
>    that MinGW uses the same convention for returning structs as
>    Cygwin, so that will actually fix a few bugs.

Agreed.

> 6. Why is the line
>
>    #define getkey getch
>
>    in xm-mingw32.h needed?

getkey doesn't exist, but getch does.

> 7. Looks like ser-mingw32.c is doing nothing.  AFAICT it shouldn't be
>    necessary.

It would be for future use.

Thanks for your help, I'll submit a new diff file.

There's a few specific bugs with MinGW versions of GDB.
I hope to "./configure && make" from CVS sources soon, and then to find
out what goes wrong sometimes.

Manu.