You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Given that the doc suggests that adding a custom hasher should be possible and the necessary interfaces offer the needed abstraction, I assume that would be a feature the project is generally interested in.
I guess the impact on non-test-code would be relatively small from what I have been able to test manually and from looking into the code base.
Right now I have no idea what the scope of test code would be. Before I dive into that (and ask for guidance as my golang-integ-test-experience is zero), I wanted to start the discussion to avoid wasted effort.
Should http.authentication.hashes modules be supported, or is this too niche of a scenario to be relevant? Is there more to be considered when this is tackled than what I stated thus far (more config, guidelines on hash formats, idk...)?
The text was updated successfully, but these errors were encountered:
marvinlwenzel
changed the title
Feature Request: scan modules for all implementations of http.authentication.hashes
Feature Request: support modules implementing http.authentication.hashes
Nov 1, 2024
FYI we used to have scrypt support but we removed it because it needed to use multiple parameters instead of having a batteries included string format like bcrypt, and doesn't have as great security properties.
Hello.
Caddy Name Spaces lists
http.authentication.hashes
as a namespace for custom implementations for hashing in use withbasic_auth
andcaddy-hash-password
.As of 2.8.4, there appears to be no check for modules implementing
http.authentication.hashes
. Just hard coded cases going "bcrypt or bust".caddy/modules/caddyhttp/caddyauth/command.go
Lines 109 to 115 in 350ad38
caddy/modules/caddyhttp/caddyauth/caddyfile.go
Lines 65 to 69 in 350ad38
The
basic_auth
Docu statescaddy-hash-password
statesThose seem misleading in the first case and false in the second.
I would like to change that.
I have started working on this to get an understanding of challenges and impact to get a proper discussion started. (So far, I was able to have both bcrypt and a custom alternative working without observing bugs. No tests, performance tests, or anything done yet. If you want to have a look: https://github.com/marvinlwenzel/caddyCrypt, my current diff master...marvinlwenzel:caddyCrypt:master and the custom hash impl https://github.com/marvinlwenzel/fvttcrypt)
Should
http.authentication.hashes
modules be supported, or is this too niche of a scenario to be relevant? Is there more to be considered when this is tackled than what I stated thus far (more config, guidelines on hash formats, idk...)?The text was updated successfully, but these errors were encountered: