Guidance re. MinGW and readline

Guidance re. MinGW and readline

[Mark Mitchell]

Chet Ramey, the readline maintainer, has reviwed my patch for GDB here:

   http://sources.redhat.com/ml/gdb-patches/2005-06/msg00067.html

Chet has accepted the patch in principle, though he's decided to take a 
slightly different approach from the one that I did, modifying rltty.c 
to use conditional compilation to determine whether or not to use any of 
the various UNIX-ish terminal variants, rather than adding an entirely 
separate rlwin32tty.c file.  Chet's version (re-write?) of the patch 
will be in readline-5.1.

Chet has also decided that the (extremely minimal) Windows termcap 
functions that I provided (like "tputs" and "tgoto") should be included 
somewhere in the application, rather than in readline itself.

So, I think that there are two paths forward:

(1) Wait for readline-5.1 to be released, incorporate it into src/, and 
then to add the minimal termcap stuff to a file in gdb/ that is only 
used on MinGW.

(2) I backport Chet's changes to rltty.c to the src/readline/ 
sourcebase, and then proceed as above.  Because we know that these 
changes will be in readline-5.1, we needn't worry overmuch about 
divergence from upstread sources.

I'd prefer the second path, even though it's more work for me, in that 
it would hopefully result in more quickly reaching the state where all 
of the MinGW support is in the FSF GDB.  However, if the GDB maintainers 
don't want to go that route, I'll just wait.  I'd very much appreciate 
advice from the GDB maintainers about which path to take.

Thanks,

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: Guidance re. MinGW and readline

[Daniel Jacobowitz]

On Sun, Jul 17, 2005 at 05:17:02PM -0700, Mark Mitchell wrote:
> (2) I backport Chet's changes to rltty.c to the src/readline/ 
> sourcebase, and then proceed as above.  Because we know that these 
> changes will be in readline-5.1, we needn't worry overmuch about 
> divergence from upstread sources.

This is the right way to do it, definitely.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Guidance re. MinGW and readline

[Christopher Faylor]

On Sun, Jul 17, 2005 at 08:23:09PM -0400, Daniel Jacobowitz wrote:
>On Sun, Jul 17, 2005 at 05:17:02PM -0700, Mark Mitchell wrote:
>> (2) I backport Chet's changes to rltty.c to the src/readline/ 
>> sourcebase, and then proceed as above.  Because we know that these 
>> changes will be in readline-5.1, we needn't worry overmuch about 
>> divergence from upstread sources.
>
>This is the right way to do it, definitely.

Ditto.

cgf

Re: Guidance re. MinGW and readline

[Eli Zaretskii]

> Date: Sun, 17 Jul 2005 17:17:02 -0700
> From: Mark Mitchell <mark@codesourcery.com>
> CC: chet.ramey@case.edu
> 
> (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
> then to add the minimal termcap stuff to a file in gdb/ that is only 
> used on MinGW.
> 
> (2) I backport Chet's changes to rltty.c to the src/readline/ 
> sourcebase, and then proceed as above.  Because we know that these 
> changes will be in readline-5.1, we needn't worry overmuch about 
> divergence from upstread sources.
> 
> I'd prefer the second path, even though it's more work for me, in that 
> it would hopefully result in more quickly reaching the state where all 
> of the MinGW support is in the FSF GDB.  However, if the GDB maintainers 
> don't want to go that route, I'll just wait.  I'd very much appreciate 
> advice from the GDB maintainers about which path to take.

If you can afford the additional effort, go the second path.

Thanks.

Re: Guidance re. MinGW and readline

[Mark Kettenis]

   Date: Sun, 17 Jul 2005 17:17:02 -0700
   From: Mark Mitchell <mark@codesourcery.com>

   (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
   then to add the minimal termcap stuff to a file in gdb/ that is only 
   used on MinGW.

   (2) I backport Chet's changes to rltty.c to the src/readline/ 
   sourcebase, and then proceed as above.  Because we know that these 
   changes will be in readline-5.1, we needn't worry overmuch about 
   divergence from upstread sources.

You'll still need the minimal termcap stuff in gdb/ in this case isn't it?

(And yes I too think that's the way to go.)

Mark

Re: Guidance re. MinGW and readline

[Mark Mitchell]

Mark Kettenis wrote:
>    Date: Sun, 17 Jul 2005 17:17:02 -0700
>    From: Mark Mitchell <mark@codesourcery.com>
> 
>    (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
>    then to add the minimal termcap stuff to a file in gdb/ that is only 
>    used on MinGW.
> 
>    (2) I backport Chet's changes to rltty.c to the src/readline/ 
>    sourcebase, and then proceed as above.  Because we know that these 
>    changes will be in readline-5.1, we needn't worry overmuch about 
>    divergence from upstread sources.
> 
> You'll still need the minimal termcap stuff in gdb/ in this case isn't it?

Correct.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: Guidance re. MinGW and readline

[Christopher Faylor]

On Mon, Jul 18, 2005 at 08:21:09AM +0200, Mark Kettenis wrote:
>   Date: Sun, 17 Jul 2005 17:17:02 -0700
>   From: Mark Mitchell
>
>   (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
>   then to add the minimal termcap stuff to a file in gdb/ that is only 
>   used on MinGW.
>
>   (2) I backport Chet's changes to rltty.c to the src/readline/ 
>   sourcebase, and then proceed as above.  Because we know that these 
>   changes will be in readline-5.1, we needn't worry overmuch about 
>   divergence from upstread sources.
>
>You'll still need the minimal termcap stuff in gdb/ in this case isn't
>it?

Actually, I wonder if libiberty would be a better place for the minimal
termcap stuff.

cgf

Re: Guidance re. MinGW and readline

[Mark Mitchell]

Christopher Faylor wrote:
> On Mon, Jul 18, 2005 at 08:21:09AM +0200, Mark Kettenis wrote:
> 
>>  Date: Sun, 17 Jul 2005 17:17:02 -0700
>>  From: Mark Mitchell
>>
>>  (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
>>  then to add the minimal termcap stuff to a file in gdb/ that is only 
>>  used on MinGW.
>>
>>  (2) I backport Chet's changes to rltty.c to the src/readline/ 
>>  sourcebase, and then proceed as above.  Because we know that these 
>>  changes will be in readline-5.1, we needn't worry overmuch about 
>>  divergence from upstread sources.
>>
>>You'll still need the minimal termcap stuff in gdb/ in this case isn't
>>it?
> 
> 
> Actually, I wonder if libiberty would be a better place for the minimal
> termcap stuff.

I knew someone would suggest that...

OK, I'll try that approach.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304

Re: Guidance re. MinGW and readline

[Ian Lance Taylor]

Christopher Faylor <me@cgf.cx> writes:

> >   (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
> >   then to add the minimal termcap stuff to a file in gdb/ that is only 
> >   used on MinGW.
> >
> >   (2) I backport Chet's changes to rltty.c to the src/readline/ 
> >   sourcebase, and then proceed as above.  Because we know that these 
> >   changes will be in readline-5.1, we needn't worry overmuch about 
> >   divergence from upstread sources.
> >
> >You'll still need the minimal termcap stuff in gdb/ in this case isn't
> >it?
> 
> Actually, I wonder if libiberty would be a better place for the minimal
> termcap stuff.

I don't see why, unless we seriously think that some program other
than gdb is going to want to use it.  If I understand the earlier
messages, the only point to the minimal termcap stuff is to use it
with readline on MinGW.  That seems fairly special purpose to me, and
not the sort of thing we usually put it into libiberty.

Ian

Re: Guidance re. MinGW and readline

[Christopher Faylor]

On Mon, Jul 18, 2005 at 09:55:31AM -0700, Ian Lance Taylor wrote:
>Christopher Faylor <me@cgf.cx> writes:
>
>> >   (1) Wait for readline-5.1 to be released, incorporate it into src/, and 
>> >   then to add the minimal termcap stuff to a file in gdb/ that is only 
>> >   used on MinGW.
>> >
>> >   (2) I backport Chet's changes to rltty.c to the src/readline/ 
>> >   sourcebase, and then proceed as above.  Because we know that these 
>> >   changes will be in readline-5.1, we needn't worry overmuch about 
>> >   divergence from upstread sources.
>> >
>> >You'll still need the minimal termcap stuff in gdb/ in this case isn't
>> >it?
>> 
>> Actually, I wonder if libiberty would be a better place for the minimal
>> termcap stuff.
>
>I don't see why, unless we seriously think that some program other
>than gdb is going to want to use it.  If I understand the earlier
>messages, the only point to the minimal termcap stuff is to use it
>with readline on MinGW.  That seems fairly special purpose to me, and
>not the sort of thing we usually put it into libiberty.

It just seemed to me that there was at least a chance that some other
GNU project might eventually want to use readline with mingw.

cgf

Re: Guidance re. MinGW and readline

[Ian Lance Taylor]

Christopher Faylor <me@cgf.cx> writes:

> >> Actually, I wonder if libiberty would be a better place for the minimal
> >> termcap stuff.
> >
> >I don't see why, unless we seriously think that some program other
> >than gdb is going to want to use it.  If I understand the earlier
> >messages, the only point to the minimal termcap stuff is to use it
> >with readline on MinGW.  That seems fairly special purpose to me, and
> >not the sort of thing we usually put it into libiberty.
> 
> It just seemed to me that there was at least a chance that some other
> GNU project might eventually want to use readline with mingw.

To me that is an argument for putting it into readline itself.
Reductively, anybody who uses readline does use readline, but does not
necessarily use libiberty.

But I don't feel all that strongly about this.  I'm just doubtful that
anybody other than gdb will ever use it out of libiberty.

Ian

Re: Guidance re. MinGW and readline

[Daniel Jacobowitz]

On Mon, Jul 18, 2005 at 10:26:38AM -0700, Ian Lance Taylor wrote:
> But I don't feel all that strongly about this.  I'm just doubtful that
> anybody other than gdb will ever use it out of libiberty.

Let's put it in GDB for now, then.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: Guidance re. MinGW and readline

[Christopher Faylor]

On Mon, Jul 18, 2005 at 01:33:16PM -0400, Daniel Jacobowitz wrote:
>On Mon, Jul 18, 2005 at 10:26:38AM -0700, Ian Lance Taylor wrote:
>> But I don't feel all that strongly about this.  I'm just doubtful that
>> anybody other than gdb will ever use it out of libiberty.
>
>Let's put it in GDB for now, then.

No objections.

cgf

Re: Guidance re. MinGW and readline

[Mark Kettenis]

   From: Ian Lance Taylor <ian@airs.com>
   Date: 18 Jul 2005 09:55:31 -0700

   Christopher Faylor <me@cgf.cx> writes:

   > Actually, I wonder if libiberty would be a better place for the minimal
   > termcap stuff.

   I don't see why, unless we seriously think that some program other
   than gdb is going to want to use it.  If I understand the earlier
   messages, the only point to the minimal termcap stuff is to use it
   with readline on MinGW.  That seems fairly special purpose to me, and
   not the sort of thing we usually put it into libiberty.

I agree with Ian here.  We should not add the additional burden of
maintining this both trees unless there actually is a benefit.

Mark