Beginners wanting to host SearXNG for use in their program won't have issues with the bot protection anymore.

I suspect that disabling the limiter will only shift the problem to the group of public instances, which will then run poorly .. which will ruin our good reputation and cause even more unnecessary bug entries in the SearXNG repo.

The problem is that many people don't read the installation instructions or at least read the comments in the settings file.

I don't understand why we should disable the link token method as well... yet another trap a public operator can fall into...

We developed the limiter to help public instances... that should remain our goal: building robust public instances... please don't lets focus on all the tinkering solutions.

public instances are already forced to have the bot protection and limiter activated: https://github.com/searxng/searxng/blob/064eb504733c250d2e43b6db8ac37812ad18d2fa/searx/limiter.py#L235-L251

Or you are talking about public instances not listed on searx.space?

Or you are talking about public instances not listed on searx.space?

Public instances in general.

If the limiter is activated, it is recommended to also activate all protective measures.

I think this is more of a documentation issue. If one wants to run a public instance not listed on https://searx.space, we should explain all the recommended steps. Those people that run public services have more ease in reading doc and following some doc. Than beginners that just want to run a private instance.

SearXNG is quite hard to install for a beginner, and I want to change this. The reason why other projects like whoogle boomed in popularity in self-hosted community is that it's simple to install: https://github.com/benbusby/whoogle-search#install-1. Only a few commands and you are go to go.

This PR is just one step towards making SearXNG easier to install locally. Other steps like dropping custom files like uwsgi.ini in Docker will also ease the local install: searxng/searxng#4820

Can we agree on not merging pull requests that still have an ongoing discussion running? What's the point of your "at least 2 approval" pr requirement if you merge it anyways even though there are critical comments against it?

It has been discussed for a long time with dalf, for which he agreed to. Back in the days, those parameters were "on" by default because we did not have "public_instance" setting and searxng-docker was mostly only for deploying public instances for being listed on https://searx.space

Now it does not make sense to keep them active by default since the target for this Docker compose has shifted towards the average user. And the average user wants to easily deploy SearXNG and be done with it. Not deploying a public instance and having issues related to bot protection.

See searxng/searxng#2832 (comment)

As I already said

we should ensure that the limiter just works without any additional setup steps necessary.

It's not fixing anything to disable an option, that would increase the average engine reliability, by default...

And even if dalf agreed at some point, Markus and I still disagreed with reasonable arguments, so obviously there's not a general agreement to merge this...