Re: [PATCH] gdb: add Python events for program space addition and removal

[Andrew Burgess <aburgess@redhat.com>]

Tom Tromey <tom@tromey.com> writes:

>>>>>> "Andrew" == Andrew Burgess via Gdb-patches <gdb-patches@sourceware.org> writes:
>
> Andrew> Initially I just wanted a Python event for when GDB removes a program
> Andrew> space, I'm writing a Python extension that caches information for each
> Andrew> program space, and need to know when I should discard entries for a
> Andrew> particular program space.
>
> Andrew> But, it seemed easy enough to also add an event for when GDB adds a
> Andrew> new program space, so I went ahead and added both new events.
>
> This approach seems totally fine to me, but it's worth noting that you
> could attach data directly to a Progspace if the tp_hash field were
> filled in.  Some other Python classes in gdb do this.  This makes caches
> easier because the cleanup is automatic.

I'm not sure I understand how tp_hash helps here?

gdb.Progspace does have a __dict__ attribute, so things like:

  progspace.my_random_var = ....

works just fine, and I did consider this.  But I wasn't sure if this was
"good practice".  My concern was mostly that this might lead to clashes
between different extensions if good names are not chosen for
'my_random_var', or even clashes between extensions and future GDB
enhancements -- my understanding is that if we later added
'my_random_var' as a builtin attribute this would take precedence over
the user defined attribute, breaking their extension.

>
> Andrew> +No @code{NewProgspaceEvent} is emitted for the very first program
> Andrew> +space, which is assigned to the first inferior.  This first program
> Andrew> +space is created within @value{GDBN} before and Python scripts are
>
> Typo, "before any"

Fixed, thanks.

>
> Other than that it looks good.

I'll likely merge this anyway, this doesn't stop folk making use of the
__dict__ approach if they prefer -- as you point out, it does have a
significant benefit in that it automates the cleanup.

>
> Approved-By: Tom Tromey <tom@tromey.com>

Thanks,
Andrew