Mingw32

Mingw32

[M.A. Henderson]

Will there be an Insight version for Mingw32?

Mingw32

[James Ingham]

M.A. Henderson writes:
 > Will there be an Insight version for Mingw32?
 > 
 > 

Gdb uses a whole bunch of Cygwin functionality to do its job on
Windows, and there are currently no plans to rewrite these sections to
call directly into the Windows API.  That would be a pretty big job,
with no clear payoff.  So, no, a Mingw32 version of Insight is not
likely to happen.

Jim

-- 
++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++
Jim Ingham                                              jingham@cygnus.com
Cygnus Solutions Inc.

Re: Mingw32

[Mumit Khan]

James Ingham <jingham@cygnus.com> writes:
> 
> Gdb uses a whole bunch of Cygwin functionality to do its job on
> Windows, and there are currently no plans to rewrite these sections to
> call directly into the Windows API.  That would be a pretty big job,
> with no clear payoff.  So, no, a Mingw32 version of Insight is not
> likely to happen.
> 

Actually, I've had a mingw port for a few years now. It requires a few
simple changes to gdb (win32-nat.c and the config files). The problem
is readline -- the older version had a "minimal" mode, but not available
anymore, and I have to essentially use a older version of readline to
get the job done. For mingw diffs to gdb-4.1[78], see:

  ftp://ftp.xraylith.wisc.edu/pub/khan/gnu-win32/mingw32/ports/

I haven't had the time to write up ChangeLog entries, but it's pretty
self-explanatory. When I get some time, I'll clean up the patches and
submit (and there's the assignment issue I keep on forgetting about). 

Unless the internals of gdb in insight is quite different and somehow
uses other Cygwin functionality, it should work.  Of course, lots of
mingw-specific tcl/tk changes are probably not in the Cygnus tree,
so that may involve some work.

Regards,
Mumit

Re: Mingw32

[James Ingham]

Mumit,

 > James Ingham <jingham@cygnus.com> writes:
 > > 
 > > Gdb uses a whole bunch of Cygwin functionality to do its job on
 > > Windows, and there are currently no plans to rewrite these sections to
 > > call directly into the Windows API.  That would be a pretty big job,
 > > with no clear payoff.  So, no, a Mingw32 version of Insight is not
 > > likely to happen.
 > > 
 > 
 > Actually, I've had a mingw port for a few years now. It requires a few
 > simple changes to gdb (win32-nat.c and the config files). The problem
 > is readline -- the older version had a "minimal" mode, but not available
 > anymore, and I have to essentially use a older version of readline to
 > get the job done. For mingw diffs to gdb-4.1[78], see:
 > 
 >   ftp://ftp.xraylith.wisc.edu/pub/khan/gnu-win32/mingw32/ports/
 > 
 > I haven't had the time to write up ChangeLog entries, but it's pretty
 > self-explanatory. When I get some time, I'll clean up the patches and
 > submit (and there's the assignment issue I keep on forgetting about). 
 > 
 > Unless the internals of gdb in insight is quite different and somehow
 > uses other Cygwin functionality, it should work.  Of course, lots of
 > mingw-specific tcl/tk changes are probably not in the Cygnus tree,
 > so that may involve some work.
 > 

This sounds cool.  How about all the socket and serial code handling
stuff needed for remote targets?  Do you get all that working as well,
or is this just for the native debugger?  I was under the impression
that Windows did not have a Posix'y interface for these functions.

Jim

-- 
++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++
Jim Ingham                                              jingham@cygnus.com
Cygnus Solutions Inc.

Re: Mingw32

[Mumit Khan]

James Ingham <jingham@cygnus.com> writes:
> 
> This sounds cool.  How about all the socket and serial code handling
> stuff needed for remote targets?  Do you get all that working as well,
> or is this just for the native debugger?  I was under the impression
> that Windows did not have a Posix'y interface for these functions.

Just a simple old native debugger, and that's why it was so easy to
do. The remote stuff will require need, time and expertise, all of 
which I lack when it comes to gdb.

Regards,
Mumit

Re: Mingw32

[James Ingham]

Mumit,

 > James Ingham <jingham@cygnus.com> writes:
 > > 
 > > This sounds cool.  How about all the socket and serial code handling
 > > stuff needed for remote targets?  Do you get all that working as well,
 > > or is this just for the native debugger?  I was under the impression
 > > that Windows did not have a Posix'y interface for these functions.
 > 
 > Just a simple old native debugger, and that's why it was so easy to
 > do. The remote stuff will require need, time and expertise, all of 
 > which I lack when it comes to gdb.
 > 

Yeah, that's what I guessed.  Converting the remote stuff would not be
all that hard, in that most of this stuff is somewhat layered out in
gdb.  On the other hand, THAT is what we mostly use Cygwin for, 'cause
it means we don't have to write Windows code in gdb to handle the
transport layer for remote debugging.  You could do it, but I guess
that I don't really see the point.  Cygwin works fine for this, and
converting to Mingw for these functions would probably just involve
ripping code out of Cygwin and stuffing it in gdb, which is not wise
for code maintainance...

Jim

-- 
++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++==++
Jim Ingham                                              jingham@cygnus.com
Cygnus Solutions Inc.

Re: Mingw32

[Mumit Khan]

James Ingham <jingham@cygnus.com> writes:
> 
> Yeah, that's what I guessed.  Converting the remote stuff would not be
> all that hard, in that most of this stuff is somewhat layered out in
> gdb.  On the other hand, THAT is what we mostly use Cygwin for, 'cause
> it means we don't have to write Windows code in gdb to handle the
> transport layer for remote debugging.  You could do it, but I guess
> that I don't really see the point.  Cygwin works fine for this, and
> converting to Mingw for these functions would probably just involve
> ripping code out of Cygwin and stuffing it in gdb, which is not wise
> for code maintainance...
> 

I agree entirely. I needed a Mingw native debugger to debug Cygwin 
startup/runtime, and starting getting requests for a native debugger
from mingw gcc who didn't want to download yet another DLL. 

Looking at my current schedule, I guess I'll simply ask mingw users
to use Cygwin port, and I can help by repackaging the whole bit in
native format and including cygwin dll.

Regards,
Mumit

Re: Mingw32

[Mumit Khan]

"M.A. Henderson" <dchender@esn.net> writes:
> Will there be an Insight version for Mingw32?

If I get some free time to add the mingw-specific stuff (mostly 
configuration, but there's bits and pieces in the tcl/tk/itcl/blt
that needs changing as well), I'll get a binary distrib together.

In the meantime, you can always use the Cygwin one. All you need
in addition is the Cygwin dll available from the Cygwin site:
  http://sourceware.cygnus.com/cygwin/
  ftp://sourceware.cygnus.com/pub/cygwin/cygwin-b20/

Put the cygwin1.dll (after decompressing) in the bin directory along
with gdb etc and you're set to go.

Regards,
Mumit