Paulgudring
commented
9 months ago In short,
- for users in need, based on discussion results in #256, try
geolocation-!cn & !gfw. It changes the least and shall conform with the repo's principle. somewhat "local blacklist in global whitelist" - add
@cnto all accessible website in category-scholar-!cn (not recommended, it breaks up the set) - add
@cnto website with isp server in china (my suggestion, pull request after finding a proven way).
Thx you all for great effort and contribution to the community
Now, the lists of
category-scholarandcategory-scholar-!cnis separated based on their geo location, which is consistent with the repo's principle. However, there are websites that can be accessed in china, no matter they have isp server in china or their overseas services are not blocked, some of which authorize content accessibility by users' ip (like clarivate).There have been issues discussing it, like #674. I agree with the maintainer's idea and I'm thoroughly aware of the difficulty. The solution there is to fall back to geoip resolve. The process is as below:
To avoid dns leak, I promote solution like
category-scholar-cn-category-scholar-!cn@cn-category-scholar-!cn. In that way, the process can be as below:For now, I use domain specific rules in my config, but it lower the performance, and readability of the config.
I'm not sure about the principle of
@cnattribute. Is the criteria strictly restricted to whether there is isp server in china?I understand the problem lies in the high specificity caused by variences between different network environment. I wrote a simple code snippet to test their accessibility (just pinged them and many of the results were wrong). But I havent figure out a proper way to test in batch connectivity across different networks so I didnt put forward a pr.
But if connectivity is the only consideration, the proposal seems to be to just assign
@cnto domains that don't appear on the gfwlist, i.e., blacklisting exemptions for academic sites within the white-listing configuration as a whole, which deviates from the repository's principles of set and my intent, so I believe the strictest standard for this attribute should be having isp server in china (like clarivate - webofscience.com).