Re: [dsdp-dd-dev] Launch conifguration for C++ applications

[Pawel Piech <pawel.piech@xxxxxxxxxxxxx>]

Hi Veenu,

Veenu Verma (AS/EAB) wrote:

Hello Pawel

>We are refactoring parts of CDT's MI
>implementation to create a pretty much brand new debug model
>implementation based on DSF.

Do you have an idea how much effort could it be to move from
CDI based GDB/MI implementation to DSF based implementation ?

This really depends on how much you will need to customize your CDI-MI implementation.  If you do not customize it at all, then theoretically, the change to DSF-MI should require no effort at all :-)  But from the context of the conversation I assume that you are making at least some extensions to the CDI-MI implementation, and these extensions will need to be ported.  My intent (and hope) is that making custom extensions to the DSF-MI implementation will be much easier than in CDI-MI, but in the end you might be a better judge of that.

>Roughly speaking
>though, our goal is to have an implementation that is functionally
>equivalent with CDI with some additional features in the areas of
>multi-core debugging and view customization.

Can you give a rough estimate on time for the release conatining MI implementation
Over DSF with functionality atleast equal to the current CDI ?

We intend to have a fully-functional DSF-MI implementation in time for the Europa release, so we'll be releasing corresponding milestones as well.  Right now we are working on the detailed plan for the feature list and on what we can have in the milestones.  We plan on presenting this plan at the CDT summit in September.  If plan on attending it, we'd be happy to answer you any questions on it then.

We (Wind River) are actually facing the same challenge, since we are going to port our commercial debugger on top of DSF-MI.  Our commercial debugger does have numerous extensions over standard GDB and we are going to have to have a way to accommodate them.

Cheers
Pawel  

Thanx
Veenu

------------------------------

Message: 3
Date: Thu, 24 Aug 2006 09:10:17 +0100
From: "Mikhail Khodjaiants" <Mikhail.Khodjaiants@xxxxxxx>
Subject: RE: [dsdp-dd-dev] Launch conifguration for C++ applications
To: "Device Debugging developer discussions" <dsdp-dd-dev@xxxxxxxxxxx>
Message-ID:
        <FDCFD1E400A1D84C930EE79B936FEC09020CA301@xxxxxxxxxxxxxxxxxx>
Content-Type: text/plain;       charset="us-ascii"

Hi,

Just as an addition to the Pawel's comment, I am currently working on the flexible hierarchy support for the CDI model. It is not going to affect the existing debugger imlementations based on CDI. At the same time it will allow to use the CDI model "in the areas of multi-core debugging and view customization".

Regards,
Mikhail Khodjaiants,
ARM Ltd.

-----Original Message-----
From: dsdp-dd-dev-bounces@xxxxxxxxxxx
[mailto:dsdp-dd-dev-bounces@xxxxxxxxxxx] On Behalf Of Pawel Piech
Sent: 23 August 2006 21:03
To: dsdp-dd-dev@xxxxxxxxxxx
Subject: Re: [dsdp-dd-dev] Launch conifguration for C++ applications

Hi Veenu,
 From the technical point of view.  We are refactoring parts of CDT's MI implementation to create a pretty much brand new debug model implementation based on DSF.  We do not plan to change the UI components of the CDT debugger implementation unless it is necessary to make them work with new APIs.  We also would like to maintain compatibility with the 3-rd party tools by implementing the platform debug model (and probably the CDT extensions to it) along side new debug model APIs.

 From the planning point of view.  We still only in the early stages of creating the MI implementation on top of DSF and we are now putting together a project plan for the Europa release.  Roughly speaking though, our goal is to have an implementation that is functionally equivalent with CDI with some additional features in the areas of multi-core debugging and view customization.  This implementation will need co-exist with the CDI based MI implementation at least until they are feature transparent.

As far as the reason for choosing one or the other, the trade off is performance and extendability in DSF vs. simplicity of design and implementation in CDI, more details are outlined in the DSF white paper <http://dsdp.eclipse.org/help/latest/topic/org.eclipse.dd.dsf.doc/docs/d

sf_white_paper.html>.

Cheers
Pawel

Veenu Verma (AS/EAB) wrote:
>
> Hello,
> A question regarding the launch configurations/delegates for Debug  in

> DSF core API's.
> When I see the source code for DSF, I see re-implementation of some
> stuff from CDT I can see that the CDT debug model is the basis but
> what about CDT implementation of GDB over MI ???
> Is the new DSF implementation going to take place of the existing CDT
> implementation(cdt.debug.mi.core) or will they co-exist  and if yes
> then
>
> What would be the reason to choose one over the other ?
>
> I am asking because we plan to implement a debugging solution using
> existing CDT API's If we write a debugging solution for our system (
> GDB over MI ) using cdt.debug.mi.core and related APIs then does it
> mean that we will have to replace it with DSF APIs in dsf.mi.core
> later Or am I not getting this whole thing properly
>
> Thanx in advance
> Veenu
>

_______________________________________________
dsdp-dd-dev mailing list
dsdp-dd-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev