< michagogo>
PRab: "bitcoin-builder" isn't a thing
< michagogo>
You need the `bitcoin` repo (specifically the contrib/gitian-descriptors and depends dirs), your fork of the `gitian.sigs` repo so you can add to it, and the `devrandom/gitian-builder` repo.
< michagogo>
PRab: are you building in KVM on a Linux machine? Or in LXC in a VM? Or something else?
< wumpus>
then in ProcessOneShots, called by ThreadOpenConnections they are handled, one by one
< jonasschnelli>
So adding a tor hidden service at seed.<provider>.<tld> (or similar) that redirects to a node running on the same system that can provide .onion addr data would be something that would be worth to achieve?
< wumpus>
yes, that avoids using an exit node to conect to your domain
< sipa>
we could just only have oneshots for seeding...
< wumpus>
e.g. blablablabla.onion:8333 -> seed.bitcoin.jonasschnelli.ch:8333 means a node with Tor can connect to the onion instead, the exit node cannot interfere
< jonasschnelli>
so we could flag dns seeds in chainparams.cpp (vSeeds) with a boolean for "tor hidden service"
< wumpus>
a flag, or an extra string field per DNS seed 'hidden service alias, if applicable'
< sipa>
or more generally, a Network enum
< sipa>
to say what network they are on
< sipa>
we could have ipv6 ones too
< wumpus>
this keeps the DNS seed data for one logical seed together
< wumpus>
right sipa
< wumpus>
although a *name* can resolve to both ipv4 and ipv6
< wumpus>
so that's already possible
< wumpus>
(we refer to the DNS seeds by name, not by IP)
< sipa>
dns seeding is annoying unflexible, because it doesn't really give ports, service bits, time last seen, ...
< sipa>
we could probably use dns seeding as a means to find oneshots
< sipa>
and the static ip list too
< jonasschnelli>
we could do filtering dns output by quering different domainnames like x.seed.<name>.<tld> ... x stands for service
< wumpus>
the reason we have NameProxy is that you cannot see from a name which network it belongs to (unless it happens to be an .onion), so it's just passed to the proxy as-is, having the proxy figure out ipv4/ipv6/...
< sipa>
yes, but it doesn't really scale well, plus we already have a protocol for getting address data in a much more featureful way
< sipa>
(the p2p protocol)
< sipa>
the big advantage of dns is that it's cached and harder tocblock
< wumpus>
yes, that was the reason to use DNS
< wumpus>
we could use more extensive data in the DNS records now that we use libevent
< wumpus>
libevent's resolver can query TXT etc
< jonasschnelli>
Yes. And i think that is a big advantage. Having a few oneShot nodes would give them the possibility to collect data, control the network.
< jonasschnelli>
Not sure if TXT records will be pass through every DNS without problems
< wumpus>
the oneshot nodes are great when DNS cannot be used, e.g. behind Tor, in which case .onion oneshot nodes would be good because the hidden service provides authentication and encryption
< jonasschnelli>
But TXT record would give us some possibilities... likes service bits, signed responses, etc.
< wumpus>
jonasschnelli: right, just raising it as an option, not saying it's the best
< wumpus>
:using DNS seeds only to find oneshot nodes" would be OK with me too
< jonasschnelli>
Someone should do a research/analysis how easy it is to deploy TXT records through the dns network.
< wumpus>
then again, bbehind a proxy you can't use the DNS seeds so you need to hardcode oneshot nodes anyhow
< sipa>
eh, right, that's independent
< wumpus>
yes
< Luke-Jr>
oneshot = connect and stay connected until it drops, right?
< Luke-Jr>
isn't that what we already do for DNS seeds?
< sipa>
Luke-Jr: connect, getaddr, disconnect
< Luke-Jr>
sipa: that seems like something new?
< sipa>
i think it's been in the code since 0.6
< wumpus>
no, that's not something new - that's our way of seeding *if* DNS cannot be used, and has been for a long time
< Luke-Jr>
hmm
< sipa>
we use it instead of dns seeding when we're behind tor
< sipa>
as it moves the dns resolving to tor
< Luke-Jr>
does it stay connected until one of the other addresses connects successfully?
< wumpus>
Luke-Jr: no, it disconnect when it receives an addr message
< fanquake>
What are we doing for signing gitian builds now? Doesn't look like the sigs have been added to the detached sigs repo?
< GitHub9>
bitcoin/master 7d325b9 Johnathan Corgan: zmq: point API link to 4.0 as that is what we are conforming to [Trivial]...
< GitHub9>
bitcoin/master a1d623d Wladimir J. van der Laan: Merge pull request #6810...
< GitHub120>
[bitcoin] laanwj closed pull request #6810: zmq: point API link to 4.0 as that is what we are conforming to (master...zmq-api-link) https://github.com/bitcoin/bitcoin/pull/6810
< jcorgan>
hmm, no idea, will have to replicate here
< cfields>
jcorgan: runs successfully with the assertion disabled
< cfields>
PRing the build fixes in just a sec
< GitHub38>
[bitcoin] theuni opened pull request #6819: WIP: force zeromq to work with static linking+mingw (master...zeromq-mingw) https://github.com/bitcoin/bitcoin/pull/6819
< GitHub8>
[bitcoin] gmaxwell closed pull request #6792: Defaults UPNP to off when discovery is disabled. (master...upnp_off_on_ip) https://github.com/bitcoin/bitcoin/pull/6792
< michagogo>
Note that you'll need to repeat that whenever there are more updates, because IIRC once you've done this process, gbuild fails if the image isn't fully up to date
< sipa>
oh, i have another openssl horror story :p
< paveljanik>
;-))
< * BlueMatt>
is now whoring out for review of #6722
< instagibbs>
related: what's the easiest way to blow up the mempool for testing purposes? My mempool is too small to test.
< instagibbs>
Missed out on the 1GB mempool train
< BlueMatt>
run testnet, get lots of coins from phantomcircuit and then just createrawtransaction in a loop creating OP_TRUE-no-signature txn?
< phantomcircuit>
instagibbs, except do it with regtest
< BlueMatt>
well , that
< BlueMatt>
but then you have to "mine" blocks
< BlueMatt>
with testnet you just make phantomcircuit give you 1000 tnbtc
< instagibbs>
hah
< instagibbs>
oh yeah, regtest fixes the block issue
< phantomcircuit>
noooo my precious coins!
< gavinandresen>
instagibbs: qa/rpc-test/smartfees.py generates a lot of transactions in -regtest mode, might be a good starting point for a mempool blowup test
< BlueMatt>
phantomcircuit: I'll buy 1 million tnbtc for $0.01
< instagibbs>
BlueMatt, that's a reset. Good job Matt.
< phantomcircuit>
damn it
< phantomcircuit>
now i have to mine them all again
< BlueMatt>
lol, testnet4 here we go
< instagibbs>
gavinandresen, thanks I'll take a look