[Bf-committers] Failing addons/module load tests

[Campbell Barton]

WRT bigger picture we could keep bpy.types for built-in types, not add
dynamically registered classes at all.

For the rare cases a script needs to access all registered classes, it
can still be done via Python's `__subclasses__`.


On Mon, Aug 28, 2017 at 6:23 PM, Campbell Barton <ideasman42 at gmail.com> wrote:
> New code in master checks for naming collisions.
> Without this check bpy.types.* isn't a reliable way to access
> registered classes.
>
> Previously duplicates were silently ignored, it mostly worked out OK
> since classes registered later never shadowed existing ones (so you
> couldn't cause access to bpy.types.Menu/Operator/Panel... etc to
> provide your own classes).
>
> On Mon, Aug 28, 2017 at 6:00 PM, Sergey Sharybin <sergey.vfx at gmail.com> wrote:
>> Hi,
>>
>> Here is situation: in current master script_load_addons
>> and script_load_modules tests are failing. Report says it's due to attempt
>> to register class twice.
>>
>> For example, in archimesh, ObjectProperties is registered from door,
>> venetian and window maker. This seems to be simple to solve: just
>> rename ObjectProperties to {Door, Venetian, Window}ObjectProperties. Is it
>> really a correct way of solving the issue?
>>
>> But then, some addons are registering operators with quite generic names,
>> for example Splint in mesh_extra tools. Shall we just rename the class?
>>
>> Other thing whicvh is failing is io_coat3d, but that one i didn't figure
>> out yet why exactly, code seems fine to me. Need someone's else eyes here.
>>
>> So., questions goes are:
>> - Are my findings are correct about root cause of the failures?
>> - Why this didn't fail before? Was it just a dangerous code which was
>> allowing to override registered classes?
>> - Shall we just rename some classes for now, so there are no conflicts?
>>
>> But bigger question is: how to avoid naming conflicts ins the future?
>> Especially for addons which are not distributed with Blender?
>>
>> --
>> With best regards, Sergey Sharybin
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> https://lists.blender.org/mailman/listinfo/bf-committers
>
>
>
> --
> - Campbell



-- 
- Campbell