< wumpus>
(e.g. I always clean up the list and have tried to remove "silly" PRs, small refactors, and self-referential documentation ones that refer to changing the release notes, and so on, but I might have removed too much or little)
< fanquake>
silly PRs :o
< wumpus>
yes, to not have a list with 1000 times "fix typo" :)
< fanquake>
sounds right, release notes are quite long as is
< wumpus>
also it's pretty easy to miscategorize things
< wumpus>
(though before you begin writing please check if the PR is actually part of this release, or an earlier one and somehow still has the tag! looks like there's a few of those)
< fanquake>
Also looks like some that we closed un-merged and still have the tag
< wumpus>
:$
< fanquake>
"TODO for release notes 0.17.0" Still "Needs release note"
< wumpus>
reminder: remove everything from that tag after -final
< wumpus>
fanquake: I think we have the INADDR_ANY one; see "The rpcallowip option can no longer be used to automatically listen " under configuration option changes
< wumpus>
updated
< fanquake>
wumpus yes good call
< nothingmuch>
what is the purpose of NET_INTERNAL address?
< nothingmuch>
(i've looked at #10466 which introduces it, but i don't understand what problem it actually solved)
< gribble>
https://github.com/bitcoin/bitcoin/issues/11512 | Use GetDesireableServiceFlags in seeds, dnsseeds, fixing static seed adding by TheBlueMatt · Pull Request #11512 · bitcoin/bitcoin · GitHub
< nothingmuch>
wait, that's not it either... sorry for spamming, too many open tabs... 10446
< dongcarl>
fanquake: Hey let me know if there are things I didn't address yet for your help on Guix! Too many threads haha. But the tool I linked to in https://github.com/fanquake/core-review/issues/5 should make it much easier to just use capabilities
< wumpus>
nothingmuch: NET_INTERNAL is used for IP ranges that are reserved for internal networks like 192.168.*
< wumpus>
nothingmuch: they're not supposed to be gossiped over addr messages
< nothingmuch>
wumpus: that's what i guessed when i first found it in the code (i was trying to figure out proxy selection logic), but IsInternal seems to specifically match an ipv6 private sub-range
< nothingmuch>
dongcarl: OH HAI!
< dongcarl>
nothingmuch: Check out `SetInternal`
< nothingmuch>
that's the part that got me confused ^_^
< wumpus>
nothingmuch: SetInternal is a scheme to generate fake internal addresses for the addrdb bucketing logic
< wumpus>
ah yes you already had #10446, there cfields kind of explains it
< dongcarl>
nothingmuch: basically, instead of having an actual IP, you have (0xFD + sha256("bitcoin")[0:5] + sha256(hostname))[0:16], which is useful in cases where you don't actually care about the IP, but the hostname, e.g. for seeds
< * dongcarl>
might be wrong
< nothingmuch>
i read through that explanation, but didn't understand the problem statement ("resolve an extra host")
< nothingmuch>
dongcarl: i think that's right, and that the goal is to get LookupHost to return an entry which would not be connected to under any circumstances
< nothingmuch>
but i'm still confused as to why that would happen
< dongcarl>
the "not be connected to under any circumstance" part is not as important as "we can identify these seeds by their hostname rather than their IP in AddrMan"
< dongcarl>
nothingmuch: does that make sense?
< nothingmuch>
hasn't clicked yet, staring hard at the call sites of SetInternal in net.cpp...
< dongcarl>
Okay, I'd suggest reading the commit message again, and lmk where you're stuck
< nothingmuch>
dongcarl: sorry, IRL distractions + thundering herd of messages... i reread the subsequent commit and it makes sense now. thanks!
< gmaxwell>
12:02:39 < mischa010> hi, compact blocks seems to reduce orphan rates by ~75%, would there happen to be a source somewhere stating this?:p
< gmaxwell>
75% sounds like a really low figure, I believe the effect is greater than that.
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #15629: init: Throw error when network specific config is ignored (master...1903-InitErrorConf) https://github.com/bitcoin/bitcoin/pull/15629
< dongcarl>
Trying to understand qt: at a high level, the user-visible options are in `OptionsModel` whereas the non-user-visible ones are kept in `gArgs`, right?
< dongcarl>
(of course there's interactions between them but... just talking in general)
< ryanofsky>
most options will be in gArgs regardless, optionsmodel.cpp will set them there
< ryanofsky>
only settings that aren't in gargs are things that are gui specific like minimize to tray
< dongcarl>
ryanofsky: I see. When we have multiprocess bitcoin, will it be an interface violation for optionsmodel.cpp to include net.h?
< ryanofsky>
for now yeah. i was kind of overzealous in #10244 and changed qt to only access gArgs indirectly through the Node interface
< ryanofsky>
i want to change that to get some simplifications, but for now it is the current state
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #15631: qa: mininode: Clearer error message on invalid magic bytes (master...1903-qaMagicMininode) https://github.com/bitcoin/bitcoin/pull/15631
< ryanofsky>
if you know what problem you want to solve, i can point you in the right direction. last thing I heard about #14856, is that you wanted to have multiple connmans in a single gui, but that didn't quite make sense to me
< mischa010>
btw measurements seem to go to zero somewhere in 2017, i wonder if they stopped collecting data or orphan rate actually went to zero :p
< gmaxwell>
mischa010: it's not zero, I assume you're looking at blockchain.info which basically gives garbage data.
< gmaxwell>
they aren't even trying.
< gmaxwell>
orphaning is caused by latency at many different places in the stack, orphaning related to non-propagation causes doesn't propagate well, and so its difficult to observe.
< gmaxwell>
if you look from only one node/geography you'll only see a tiny fraction of it.
< gmaxwell>
for some reason bc.i doesn't even show that much, I'd guess their collection is just broken entirely.
< mischa010>
thanks, it's the best data i could find
< mischa010>
so nice to have a list of reasons to explain divergence of my results