< bitcoin-git>
bitcoin/master 6d3689f Vasil Dimov: sync: print proper lock order location when double lock is detected
< bitcoin-git>
bitcoin/master a21dc46 Vasil Dimov: sync: const-qualify the argument of double_lock_detected()
< bitcoin-git>
bitcoin/master db058ef Vasil Dimov: sync: use HasReason() in double lock tests
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #20507: sync: print proper lock order location when double lock is detected (master...double_lock_print_location) https://github.com/bitcoin/bitcoin/pull/20507
< promag>
vasild: could be, thats a valid use case I guess - point -rpccookiefile to a shared location
< bitcoin-git>
[bitcoin] hebasto opened pull request #20544: build: Do not repeat warning names in -Werror=... options (master...201202-werror) https://github.com/bitcoin/bitcoin/pull/20544
< sipa>
if only socket/fifo files on network filesystems also allowed network communication between multiple mounting systems
< MarcoFalke>
[06:41] <fanquake> Having to double-click through to see in-progress cirrus runs is very annoying
< MarcoFalke>
Agree
< MarcoFalke>
Thats a GitHub bug, though. bitcoinbuilds is also affected.
< MarcoFalke>
Core doesn't support Knots, it never has been
< luke-jr>
MarcoFalke: this is about forward compatbility
< MarcoFalke>
When it comes to auth, we should be strict in validating it. Not silently accept trash and return true. This might lead to loss of funds or confusion at best.
< luke-jr>
how?
< luke-jr>
ignoring it at startup is a safe failure mode; it is still enforced overly strict at runtime
< MarcoFalke>
Ok fine, it doesn't return true
< MarcoFalke>
Though silently ignoring invalid auths is just asking for hard-to-debug issues
< luke-jr>
has that *ever* been the case?
< MarcoFalke>
yes, args not being validated has made people to create nonsense config files in the past
< luke-jr>
but that's not what we were discussing..
< luke-jr>
has ignoring invalid rpcauth lines ever created a situation where debugging was confusing or difficult?
< MarcoFalke>
not for me personally. I can't speak about other users. I like the change, but I am not married to it, if you feel strongly, feel free to revert it. This is a minor change to user experience, not something for me to have a lenghty discussion about.
< promag>
flawed conceptually is too much no? it validates by the current format and forces the user to fix it. normal stuff
< MarcoFalke>
the only difference is that the user will see a permission error on startup and be asked to fix it as opposed to later when the server is running and the permission in denied and they'd have to restart the server
< promag>
you can workaround your case by splitting the config in multiple files which can share some of the config
< bitcoin-git>
[bitcoin] luke-jr closed pull request #12674: RPC: Support addnode onetry without making the connection priviliged (master...rpc_onetry_nonpriv) https://github.com/bitcoin/bitcoin/pull/12674
< bitcoin-git>
[bitcoin] luke-jr opened pull request #20551: RPC/Net: Allow changing the connection_type for addnode onetry (master...rpc_onetry_conntype) https://github.com/bitcoin/bitcoin/pull/20551