Alexander Sosedkin created an issue: 
https://gitlab.com/gnutls/gnutls/-/issues/1564



!1550 has introduced, and !1583 has extended `--strict-x509`, but, without 
being enabled by default, the compliance improvements offered by the switch lie 
dormant.
On the other hand, enabling the option comes at an interoperability cost, so 
packagers are discouraged to enabled it, as when users encounter a 
non-compliant certificate and come to them, their workarounding options are 
currently limited to recompiling gnutls only. Introducing some form of a 
runtime switch might help them make the leap.

A runtime switch can come in several varieties. In the most common scenario, 
where a client cannot connect to a server and has no control over the 
non-compliant certificate it's using, the ideal override would be per-host, and 
the second best override would be per-invocation, followed by whole-system. 
Per-app switching, when a specific app wants an override through the API, feels 
like the least likely to be handy. With that in mind, here's my subjective 
rating of what the switch could be:

1. environment variable. Pros: per-invocation. Cons: SUID binaries might ignore 
it. (not sure how significant is that)
2. priority string keyword. Pros: some apps allow configuring it per-invocation 
or per-app. Cons: not easy to enable system-wide with allowlisting config alone 
(not sure how significant is that)
3. configuration file directive. Pros: can be per-invocation if pointed at a 
config with an envvar. Cons: folks rocking no config (e.g. Debian) will have to 
figure out creating one

Another question is, when can the default for the compile switch be flipped. 
Next second version bump after a runtime switch is there?

Overall, what's the plan to proceed with this one?

-- 
Reply to this email directly or view it on GitLab: 
https://gitlab.com/gnutls/gnutls/-/issues/1564
You're receiving this email because of your account on gitlab.com.


_______________________________________________
Gnutls-devel mailing list
[email protected]
http://lists.gnupg.org/mailman/listinfo/gnutls-devel

Reply via email to