Me: We all know PGP is bad because it forces you to support legacy bullshit like 1024-bit RSA keys, but I'm specifically going to write a blog post about email encryption as not just a technical challenge.
-
Me: We all know PGP is bad because it forces you to support legacy bullshit like 1024-bit RSA keys, but I'm specifically going to write a blog post about email encryption as not just a technical challenge.
@delta: subtoots my blog post
Their source code: Disables TLS security to support legacy 1024-bit RSA.
-
Me: We all know PGP is bad because it forces you to support legacy bullshit like 1024-bit RSA keys, but I'm specifically going to write a blog post about email encryption as not just a technical challenge.
@delta: subtoots my blog post
Their source code: Disables TLS security to support legacy 1024-bit RSA.
@soatok please read more carefully the source code. You are in the "else" branch of "if strict_tls" and strict_tls is True by default.
It's pretty deep down in advanced menus to opt out of the default strict tls handling (with RustTLS btw). Opt out is needed for some users who would otherwise have no TLS at all in their national environment, and where they often use public WIFIs, so eavesdropping cleartext from people around is trivial but cracking RSA1024 is not.
-
@soatok please read more carefully the source code. You are in the "else" branch of "if strict_tls" and strict_tls is True by default.
It's pretty deep down in advanced menus to opt out of the default strict tls handling (with RustTLS btw). Opt out is needed for some users who would otherwise have no TLS at all in their national environment, and where they often use public WIFIs, so eavesdropping cleartext from people around is trivial but cracking RSA1024 is not.
@delta No, I understood the context. I just think it's hilarious that you think you're the exception when your code has a very clear example of the kind of shitty hacks needed for backwards compatibility.
-
@delta No, I understood the context. I just think it's hilarious that you think you're the exception when your code has a very clear example of the kind of shitty hacks needed for backwards compatibility.
@soatok Meanwhile, Signal having no fallback to plaintext. And I would genuinely be *extremely* surprised if it has a fallback that allows using 1024-bit RSA in a mode where invalid certificates and incorrect certificate hostnames are accepted…
Almost makes me curious if there is any way whatsoever to trick a user into having not strict_tls there. Not curious enough to go digging through the code, though.
-
R AodeRelay shared this topic
