Pedro Brochado

> 2. Is there a way to completely disable the check up of environmental variable "DYNACONF_" when different envvar_prefix is like below? There is also a workaround. [This example](https://www.dynaconf.com/advanced/#prefix-filtering) from...

> Is there a way to disable deep merging lists (instead, override and just keep latest list), but keeping the behavior for objects? @jgod I believe there is not. Maybe...

> I tried to put dynaconf_merge_unique as a key in the yaml file under deep_list but it doesn't seem to dedup for me. @zexuan-zhou Currently (`v3.1.12`) this works as described...

> `cuda_visible_devices: "@jinja {% for i in range(0, this.training.gpu_resource_config.n_gpus) %} {{i}} {% endfor %} | join(',')"` > gives me `"cuda_visible_devices": " 0 1 2 3 4 5 6 7 |...

I think the main doc change needed here is adding [these examples](https://github.com/dynaconf/dynaconf/issues/644#issuecomment-902313800) about merging. I've reviewed the [Merging](https://www.dynaconf.com/merging/) page to see where I could fit those examples, and they could...

> We can add validate=True to methods update, set, load_file. > > All the methods that can be called on-demand, so it is optional. It would be good to add...

Closed by #900

Closed by #907

I would advocate for using `DYNACONF_DATA__a__0__key__subkey`, or `DYNACONF_DATA__a__i0__key__subkey` (as suggested by Bruno in another occasion). In this context, `__` is an unambiguous separator, so `__a_0___` look like `a_0`. And it...

Hello @bacalfa . It is really hard to tell what is going on without being able to reproduce it, but I really don't think dynaconf can load from the wrong...