As of July 16, iocaine is set to block requests made with user agents that specify a Chrome/Chromium version up to 148. This was an extended stable channel, so Vivaldi (and presumably other Chromium-based browsers) were using this for a while. Nixpkgs unstable only updated to a Vivaldi 8.1.4087.48, which is based on Chromium 150, on July 9, and 26.05 only updated on July 11.
This created an annoying situation where I updated my config a few days before this, then a week and a half later I found that hydra.nixos.org was unusable. It's fixed now that I have updated to a more recent version, but I'm concerned that it might happen again given how strict the filter is being set (e.g. if Vivaldi is slow to update to Chromium 151, and the filter is updated to 150).
I could (and probably should) update my browser more frequently, or maybe just fake my user agent, but the filter still seems way too strict. It's simultaneously very annoying for normal users while doing basically nothing to stop scraping. Regularly updating user agents and detecting Markov chain-generated text are all trivial for someone running a large scraping operation, but the average user who updates their browser once a month has no clue about any of this and will just assume the site is dead.
Can the browser version checks be removed? A Captcha or a PoW challenge like Anubis would work better, but at least that would reduce false positives a lot.
As of July 16, iocaine is set to block requests made with user agents that specify a Chrome/Chromium version up to 148. This was an extended stable channel, so Vivaldi (and presumably other Chromium-based browsers) were using this for a while. Nixpkgs unstable only updated to a Vivaldi 8.1.4087.48, which is based on Chromium 150, on July 9, and 26.05 only updated on July 11.
This created an annoying situation where I updated my config a few days before this, then a week and a half later I found that hydra.nixos.org was unusable. It's fixed now that I have updated to a more recent version, but I'm concerned that it might happen again given how strict the filter is being set (e.g. if Vivaldi is slow to update to Chromium 151, and the filter is updated to 150).
I could (and probably should) update my browser more frequently, or maybe just fake my user agent, but the filter still seems way too strict. It's simultaneously very annoying for normal users while doing basically nothing to stop scraping. Regularly updating user agents and detecting Markov chain-generated text are all trivial for someone running a large scraping operation, but the average user who updates their browser once a month has no clue about any of this and will just assume the site is dead.
Can the browser version checks be removed? A Captcha or a PoW challenge like Anubis would work better, but at least that would reduce false positives a lot.