Re: [PATCH v1] py-gdb-readline: replace deprecated interfaces in GdbRemoveReadlineFinder

On 8/5/26 3:17 PM, Tom de Vries wrote:
> On 1/13/26 5:25 PM, Tom Tromey wrote:
>>>>>>> "Matthieu" == Matthieu Longo <[email protected]> writes:
>>
>> Matthieu> A previous patch [1] enabled readline in Python in a GDB- 
>> specific way
>> Matthieu> and blocked the standard Python readline module to prevent 
>> conflicts
>> Matthieu> with GDB by adding a custom importer raising an exception 
>> for the readline
>> Matthieu> module.
>>
>> I wonder if gdb could do better here.  To be clear, not your problem.
>>
>> Matthieu> This patch uses Patryk Sondej's approach detailed in bug 
>> 32473, but with
>> Matthieu> a slight variation regarding the finder insertion in 
>> sys.meta_path.
>> Matthieu> It also adds a new test to prevent future regression.
>>
>> Seems reasonable to me, though this isn't an area I know very well.
>>
>> Approved-By: Tom Tromey <[email protected]>
> 
> I reproduced this regression on the gdb-17-branch with python 3.6 (and 
> later as well), and can confirm that backporting the patch to the 
> gdb-17-branch fixes it.
> 
> I'm not able to test with python 3.4 though, otherwise I would apply it 
> to the branch.
> 

I remembered I had a leap 42.3 container, with system python 3.4.  It 
has gcc 4.8.5, so I can't build gdb-17-branch, so instead I build 
gdb-14-branch.  The problem reproduces, and backporting the patch 
(applies cleanly) fixes it.

So now that I know that I'm not breaking something for python 3.4, I'll 
proceed with the backport.

Thanks,
- Tom

> Thanks,
> - Tom
> 
>>
>> Tom
>