Re: [PATCH] Display full file path in MI style disassembly listing

["André Pönitz" <andre.poenitz@mathematik.tu-chemnitz.de>]

On Thu, Oct 18, 2012 at 11:17:22AM +0100, Pedro Alves wrote:
> On 10/18/2012 10:48 AM, Andrew Burgess wrote:
> 
> >>> I think we should
> >>> output a "fullname" field for MI, like we do for breakpoints.
> > 
> > I would be happy to take this approach as a compromise.
> 
> IMO, it's not really a compromise.  It's all about giving all the data
> to the frontend, so it can do whatever it pleases.
> 
>  - the file path as recorded in the compilation.  This is useful to have
>    so the user can identify issues with fullpath mappings.
>  - the file path gdb thinks that maps to in the filesystem, using the current
>    settings of "directory/set directories/etc".

I'd personally also call it a proper solution, not a compromise.

> I'd even argue that the backtrace path in question should not change the "file"
> field in MI, but instead it would add a new field to the output, if that's
> also useful for MI.

From an MI consumer's point of view anything that provides the raw data and 
a "best effort interpretation" at the same time is perfect. Not providing
"raw data" leaves no room for fixups in case the "cooked" version is "wrong",
not providing the "cooked" version leaves the consumer with re-implementing
the "cooking", often enough without access to helpful auxilliary infomation.

Andre'