Execute code in python pretty printers

Execute code in python pretty printers

[Niko Sams]

Hi,

I'm writing pretty printers for various Qt classes.
For QModelIndex I'd like to show the parent QModelIndex as child. The
parent isn't
saved in the QModelIndex object direclty - instead QModelIndex asks
the model for it's parent.
Basically it returns this->model->parent(this).
QAbstractItemModel::parent is a virtual function implemented
differently in every model.
So to find out the parent the method has to be executed.

Is this possible?

Niko

Re: Execute code in python pretty printers

[Tom Tromey]

>>>>> "Niko" == Niko Sams <niko.sams@gmail.com> writes:

Niko> For QModelIndex I'd like to show the parent QModelIndex as
Niko> child.

This reads very strangely :)

Niko> So to find out the parent the method has to be executed.
Niko> Is this possible?

No, we have not implemented inferior function calls on Value yet.

Note that even when it does work, it is not the friendliest thing to
do.  Doing this prevents your pretty-printers from working with core
files.

Tom

Re: Execute code in python pretty printers

[Niko Sams]

On Mon, Jun 29, 2009 at 19:07, Tom Tromey<tromey@redhat.com> wrote:
>>>>>> "Niko" == Niko Sams <niko.sams@gmail.com> writes:
>
> Niko> For QModelIndex I'd like to show the parent QModelIndex as
> Niko> child.
>
> This reads very strangely :)
>
> Niko> So to find out the parent the method has to be executed.
> Niko> Is this possible?
>
> No, we have not implemented inferior function calls on Value yet.
ok, at least this means it should be possible...
Is this planned for 7.0?

> Note that even when it does work, it is not the friendliest thing to
> do.  Doing this prevents your pretty-printers from working with core
> files.
I see. And I guess there is no way around it.
But better it works for normal debug session than never.

Niko

Re: Execute code in python pretty printers

[Paul Pluzhnikov]

On Mon, Jun 29, 2009 at 10:07 AM, Tom Tromey<tromey@redhat.com> wrote:

> Niko> So to find out the parent the method has to be executed.
> Niko> Is this possible?
>
> No, we have not implemented inferior function calls on Value yet.

But I believe it *is* possible (if ugly):
http://sourceware.org/ml/archer/2009-q2/msg00197.html


-- 
Paul Pluzhnikov

Re: Execute code in python pretty printers

[Tom Tromey]

>>>>> "Niko" == Niko Sams <niko.sams@gmail.com> writes:

Tom> No, we have not implemented inferior function calls on Value yet.

Niko> ok, at least this means it should be possible...
Niko> Is this planned for 7.0?

I don't have any concrete plan to implement it.
I don't know whether anybody else is working on it at the moment.

Tom

Re: Execute code in python pretty printers

[Tom Tromey]

>>>>> "Paul" == Paul Pluzhnikov <ppluzhnikov@google.com> writes:

Paul> On Mon, Jun 29, 2009 at 10:07 AM, Tom Tromey<tromey@redhat.com> wrote:
Niko> So to find out the parent the method has to be executed.
Niko> Is this possible?
>> 
>> No, we have not implemented inferior function calls on Value yet.

Paul> But I believe it *is* possible (if ugly):
Paul> http://sourceware.org/ml/archer/2009-q2/msg00197.html

This reminds me -- I'm curious about the crashes you were seeing.
Could you file a bug report?

I wonder whether these make will it difficult to implement the feature
in a real way later one.

Tom

Re: Execute code in python pretty printers

[Niko Sams]

On Mon, Jun 29, 2009 at 19:50, Paul Pluzhnikov<ppluzhnikov@google.com> wrote:
> On Mon, Jun 29, 2009 at 10:07 AM, Tom Tromey<tromey@redhat.com> wrote:
>
>> Niko> So to find out the parent the method has to be executed.
>> Niko> Is this possible?
>>
>> No, we have not implemented inferior function calls on Value yet.
>
> But I believe it *is* possible (if ugly):
> http://sourceware.org/ml/archer/2009-q2/msg00197.html
hmm... that doesn't work for me:
NameError: global name 'parse_and_eval' is not defined

Niko

Re: Execute code in python pretty printers

[Paul Pluzhnikov]

On Mon, Jun 29, 2009 at 12:12 PM, Niko Sams<niko.sams@gmail.com> wrote:

>> http://sourceware.org/ml/archer/2009-q2/msg00197.html
> hmm... that doesn't work for me:
> NameError: global name 'parse_and_eval' is not defined

Right. You need archer-tromey-python branch, or this not-yet-committed
patch to mainline:

http://sourceware.org/ml/gdb-patches/2009-06/msg00326.html


-- 
Paul Pluzhnikov

Re: Execute code in python pretty printers

[Paul Pluzhnikov]

On Mon, Jun 29, 2009 at 11:25 AM, Tom Tromey<tromey@redhat.com> wrote:

> This reminds me -- I'm curious about the crashes you were seeing.
> Could you file a bug report?

http://sourceware.org/bugzilla/show_bug.cgi?id=10344

-- 
Paul Pluzhnikov