Worried about LLM-written modules

I believe that @kofa’s OP was not concerned with slop, which is what you are reporting here. IIUC, the concern is that more people are being empowered with the ability to contribute meaningfully to the platform, and he is trying to brainstorm strategies that can be implemented so that these contributions do not become overwhelming for the core dev team. So, it’s not about undesirable contributions, but rather what steps can be taken so that the dev team can effectively cope with a higher volume of desirable contributions.

Auto-generated, pointless slop is related, but it’s a completely different problem, as discussed at length in this other post.

The keyword in the passage aboive is new contributors. The problem is not that there can be more contributions, the problem is that many of those can, potentially, come from completely outsider, untrusted contributors. A trust system needs to be in place, which is already the case by the way.

So in the end I am not sure that a lot needs to change. Some permissions / processes may need a bit of tightening here and there to reduce the amount of unwanted overhead. And devs can also start using some of this tech (maybe some do, already?) to streamline the work on their side. But overall I think that it will work just fine.

Not only that, I think that the net result will be a lot of user facing polish and useful features in darktable. Not having to spend huge amounts of time on coding details and day-to-day maintainance tasks (both of which are, for 99%, grudge work) will free a lot of dev cycles that can be spent on more rewarding “big-picture” tasks.

2 Likes