geolocation-!cn: include more sub lists

[MkQtS]

They are part of Chinese entities but exclusively serving for non-cn area

As #3198 moved them out from category-xxx-cn/geolocation-cn, this adds them into category-xxx-!cn/geolocation-!cn

If there's no objection, I'll merge this in a week

[nekolsd]

What are we doing here bro, the category-companies list states that

This list contains Tech & Cyber companies and organizations outside China mainland.

Adding domains of China mainland-based companies that serve non-cn area into this list doesn't make sense. These domains still belong to Chinese companies, not foreign ones — this contradicts the list's own description and makes the categorization misleading.

This probably goes back to how the project categorized things in the first place — some lists may have been poorly named from the start. But that's not a reason to make it worse by adding more mismatched entries.

don't do this bro

[MkQtS]

Adding domains of China mainland-based companies that serve non-cn area into this list doesn't make sense

Well, I thought these domains are used by their foreign subsidiary which may considered as non-cn companies. geolocation-!cn is a bit too generic, and these lists match the Tech & Cyber companies topic, so I added them to category-companies.

I can add them directly in geolocation-!cn, or create a new list like category-non-domestic-cn-companies(the name may be shorter...). That's not a problem.

Beyond this naming/categorization detail, actually, I am more concerned about whether this change is reasonable/necessary. I think removing @!cn domain rules out from category-xxx-cn is quite reasonable. But after this, I found the change appears to diminish the value of these removed domains(originally cn@!cn domains) as they were removed from big categories but not added to new ones. I had a feeling that since I was the one who drove them away, it should be my responsibility to find them a place to go. So it may be a choice to add them into geolocation-!cn families.

TBH, I don't like the idea to add all @cn domains to geolocation-cn and all @!cn to geolocation-!cn. It doesn't make sense to treat google@cn domains as part of cn domains. Similarly, I'm puzzled about the necessity of adding cn@!cn into -!cn lists.

Do you have any suggestions about that? Feel free to share your ideas.

[MkQtS]

How about handling it this way? @nekolsd

[Meo597]

Under no circumstances should Chinese entities be included in geolocation-!cn, as they very likely possess both domestic and international IPs.
Once you do this, it will break a large number of existing routing policies.

[MkQtS]

Once you do this, it will break a large number of existing routing policies.

Could you explain this? I don't quite understand.

In my understanding, we mark one domain with @!cn only if the domain resolves IPs outside cn, or the service is exclusively serving for non-cn area. If any domain belongs to a Chinese entity and resolves to cn IP/serves for cn area, then it should not be marked with @!cn.

Actually, some downstream projects, Loyalsoldier/MetacubeX/SagerNet, are even more aggressive. They already added cn@!cn to geolocation-!cn and !cn@cn to geolocation-cn. I didn't see much dissatisfaction from users.

Currently, this repo has moved cn@!cn out from geolocation-cn and category-xxx-cn. This PR is going to add these cn@!cn rules to geolocation-!cn. No one expressed a strong preference for this change, including myself, I have no intention to merge it now. I'd like to keep this PR open to receive more opinions.

[Meo597]

Usually, people consider geolocation-!cn and cn to represent non-mainland and mainland entities, respectively.
Therefore, loc-!cn is often seen as being used by the GFW list, especially when users have little VPS traffic or poor router.

Once overseas businesses of Chinese entities are merged into loc-!cn, it might change user intent, as they did not originally want these domains to pass through the proxy.

[MkQtS]

loc-!cn is often seen as being used by the GFW list

geolocation-!cn is not GFW list. There are many cn available domains, like apple, microsoft, google@cn. It is unreasonable to expect geolocation-!cn to achieve the same effect as GFW list.

it might change user intent, as they did not originally want these domains to pass through the proxy.

If anyone does not want to proxy all !cn traffic, then he should not proxy the whole geolocation-!cn.

I think we should do things right, not cater to existing improper perceptions/expectations.

[Meo597]

This will trigger large-scale configuration changes downstream.

Because some people want Chinese entities to connect directly via Chinese IPs, even for overseas services, especially for things like online banking.
Now they have to relearn the new semantics and adjust the order of the rules.

Placing loc-!cn@cn into cn does not have this effect; we generally trust the moral standards of international entities.