RE: GDB Annotations : Support plan

RE: GDB Annotations : Support plan

[Vladimir Prus]

Jain, Rohit wrote:

> Hi,
> 
>  
> 
> GDB is now supporting GDB/MI interface, and the annotation mechanism has
> now become obsolete.
> 
>  
> 
> Is there any plan to completely discontinue the support for annotations
> in future?
> 
> If yes, and if you know approximate release period that annotations are
> going to exist, please let me know.
> 
> If no, then is it safe to assume that GDB/MI interface and annotations
> (although obsolete!) are going to co-exist?

If I understand correctly, right now annotations are essentially fossilized 
--- nobody adds new annotations, and due to small test coverage, it's hard
to make sure that existing annotations are not broken.

IMO, it would be great to remove them, tomorrow. However, it's probably
more of a political question, than a technical one, so no estimate
of low long they will stay can be provided.

- Volodya

GDB Annotations : support plan

[Jain, Rohit]

Hi,

GDB is now supporting GDB/MI interface, and the annotation mechanism has
now become obsolete.

Is there any plan to completely discontinue the support for annotations
in future?
If yes, and if you know approximate release period that annotations are
going to exist, please let me know.
If no, then is it safe to assume that GDB/MI interface and annotations
(although obsolete!) are going to co-exist?

Regards,
Rohit

Re: GDB Annotations : support plan

[Nick Roberts]

Jain, Rohit writes:
 > Hi,
 > 
 > GDB is now supporting GDB/MI interface, and the annotation mechanism has
 > now become obsolete.
 > 
 > Is there any plan to completely discontinue the support for annotations
 > in future?

Yes.

 > If yes, and if you know approximate release period that annotations are
 > going to exist, please let me know.

AFAIK here is no timescale.

 > If no, then is it safe to assume that GDB/MI interface and annotations
 > (although obsolete!) are going to co-exist?

Building frontends that use annotations is actively discouraged.  If
GDB/MI does not currently meet all your requirements then please contribute
to its development.

Could you please post your interest in these matters here, so that we can
consider you when we make changes to to GDB/MI and annotations.

-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: GDB Annotations : support plan

[Bob Rossi]

On Wed, Jan 30, 2008 at 12:40:10PM +1300, Nick Roberts wrote:
> Jain, Rohit writes:
>  > Hi,
>  > 
>  > GDB is now supporting GDB/MI interface, and the annotation mechanism has
>  > now become obsolete.
>  > 
>  > Is there any plan to completely discontinue the support for annotations
>  > in future?
> 
> Yes.

Agreed.

>  > If yes, and if you know approximate release period that annotations are
>  > going to exist, please let me know.
> 
> AFAIK here is no timescale.

Hopefully not to soon. I still use them in my front end, but will
hopefully in 2008 switch to MI.

>  > If no, then is it safe to assume that GDB/MI interface and annotations
>  > (although obsolete!) are going to co-exist?
> 
> Building frontends that use annotations is actively discouraged.  If
> GDB/MI does not currently meet all your requirements then please contribute
> to its development.

I have to say, it would be a really bad idea to use the annotations.
Really bad. It's hard to do much of anything with what they provide.

However, GDB/MI will waste a lot of your time if your requirements
aren't meet through what it provides already. I'm just warning you.

Someone in the GDB development area basically decided to create GDB/MI 
as a nice protocol for GDB front ends. However, they didn't write a 
client to demostrate it's purposes. Each client reimplements all the 
work necessary to parse and interpret the protocol. Furthermore, every 
client wants something different and so has different expectations of 
what the interface should provide. Finally, if it doesn't do something 
you need, the standard answer is - do it yourself. Just for the icing 
on the cake, don't expect your patch to get reviewed for 6 months.

This is my experience. I have a bison parser that parses the GDB/MI
protocol, however, it only works with a bison version in CVS. Contact
me if you care.

Bob Rossi

RE: GDB Annotations : support plan

[Jain, Rohit]

Thanks for your responses, Nick and Bob.

We already have integration with Gdb using annotations right now.
But, we are considering moving to using either GDB/MI interface or Gdbtk
for better and reliable integration. Knowing ahead the Gdb's support
plan for annotation will help us making a decision.

Regards,
Rohit

-----Original Message-----
From: Bob Rossi [mailto:bob_rossi@cox.net] 
Sent: Tuesday, January 29, 2008 4:21 PM
To: Nick Roberts
Cc: Jain, Rohit; gdb@sources.redhat.com
Subject: Re: GDB Annotations : support plan

On Wed, Jan 30, 2008 at 12:40:10PM +1300, Nick Roberts wrote:
> Jain, Rohit writes:
>  > Hi,
>  > 
>  > GDB is now supporting GDB/MI interface, and the annotation
mechanism has
>  > now become obsolete.
>  > 
>  > Is there any plan to completely discontinue the support for
annotations
>  > in future?
> 
> Yes.

Agreed.

>  > If yes, and if you know approximate release period that annotations
are
>  > going to exist, please let me know.
> 
> AFAIK here is no timescale.

Hopefully not to soon. I still use them in my front end, but will
hopefully in 2008 switch to MI.

>  > If no, then is it safe to assume that GDB/MI interface and
annotations
>  > (although obsolete!) are going to co-exist?
> 
> Building frontends that use annotations is actively discouraged.  If
> GDB/MI does not currently meet all your requirements then please
contribute
> to its development.

I have to say, it would be a really bad idea to use the annotations.
Really bad. It's hard to do much of anything with what they provide.

However, GDB/MI will waste a lot of your time if your requirements
aren't meet through what it provides already. I'm just warning you.

Someone in the GDB development area basically decided to create GDB/MI 
as a nice protocol for GDB front ends. However, they didn't write a 
client to demostrate it's purposes. Each client reimplements all the 
work necessary to parse and interpret the protocol. Furthermore, every 
client wants something different and so has different expectations of 
what the interface should provide. Finally, if it doesn't do something 
you need, the standard answer is - do it yourself. Just for the icing 
on the cake, don't expect your patch to get reviewed for 6 months.

This is my experience. I have a bison parser that parses the GDB/MI
protocol, however, it only works with a bison version in CVS. Contact
me if you care.

Bob Rossi