Skip to main content


and in case you missed it: with the new addition of --ech, #curl now supports 259 command line options
#curl
in reply to daniel:// stenberg://

uh, congrats? :)

this does bring to mind the internal Sun april fools memo detailing the formation of a new Sun division to support options to "ls"...

in reply to daniel:// stenberg://

Because 260 are too many :⁠-⁠)
This entry was edited (2 weeks ago)
in reply to daniel:// stenberg://

now close you eyes, and recite them all in alphabetical order.
in reply to Tom

... it would not even be hard since curl can do any amount of transfers so it's just a matter of adding more to a command line.
in reply to daniel:// stenberg://

I just checked and "curl -abcdefghijklmnopqrstuvwxyz example.com" looks like the same as simple "curl example.com" :)

EDIT: some of them takes the other arguments as arguments, e.g. "curl -zyxwvutsrqponmlkjihgfedcba" print some argument warnings.

This entry was edited (2 weeks ago)
in reply to daniel:// stenberg://

and what is your opinion on ECH?

Do you not find it funny that the orgs that can effectively use it are the ones that disabled domain fronting? Cause for a small party one can guess which of the few services might be behind the cert offering ECH, with a big shared service it indeed becomes a larger guess (wolf hiding amongst sheep); the big service though does get to see what is the real target. And if one can observe the client's DNS lookups, one already has the information....

in reply to Jeroen Massar

@jeroen true, but domain fronting was a hit and miss hack, this is an established protocol. We also have encrypted DNS solutions these days that should prevent easy snooping there.

From what I hear quite a few networks/orgs already block ECH which could perhaps be seen as a sign that it actually might work...

in reply to daniel:// stenberg://

@jeroen ECH does not only hide the domain name. It hides lots of metadata like the ALPN or the initial parameters of QUIC, etc. It is useful even when domain fronting is not.

daniel:// stenberg:// reshared this.

in reply to Rich Felker

@dalias far away from that at this point. There is a lot to work out, and again there is a problem with lack of support in TLS libraries...
in reply to daniel:// stenberg://

Oh! Encrypted Client Hello

Fantastic, and thank you for what you do