< promag>
an alternative to force the build is to redeliver the webhook, however there's no official API for this. But I think it should be possible by using GET https://github.com/hooks/$HOOK_ID/deliveries and POST https://github.com/hooks/$HOOK_ID/deliveries/$DELIVERY_ID/redeliver with the right headers
< luke-jr>
achow101: the travis restart thing won't?
< achow101>
restarting a job will just use the same commit for that job
< achow101>
you'd need to trigger a new build for the pr, which isn't possible at least from the web ui
< luke-jr>
what if we modify our travis scripts to fetch master explicitly?
< gmaxwell>
luke-jr: re your list post, "If that is inacceptable, pick as internal key a point with unknown discrete logarithm." there is no reason your forced to have a everyone agress bypass.
< tryphe>
i noticed there's not a lot of usage of 'const' in a lot of bitcoin code where there could be. is this for readability purposes, or are changes welcome?
< sipa>
tryphe: we reject patches that only improve coding stylr
< sipa>
*style
< tryphe>
sipa, i was thinking of an improvement like so: say you pass a pointer to a thread, and you expect the pointer to be accessed non-const but the address value to not be modified. now you could enforce that with coverage tests and stuff.
< tryphe>
gah, thread/function*
< tryphe>
whereas if you don't enforce it, you won't know if you accidentally modified a value
< sipa>
yes, improving that is great, but it is very low priority noise compared to other potential changes
< sipa>
and review time is limited and precious
< tryphe>
true, but in terms of undefined behavior i think it could be helpful
< gwillen>
adding "const" feels like it should be slightly different from "mere style" changes, since it's functional rather than aesthetic, and any instance the compiler accepts is almost surely correct
< gwillen>
(so review effort ought to be limited)
< sipa>
if you actually find an issue, for example by adding const where you expect it to work and it doesn't, it is different
< tryphe>
(i realized it's impractical to change this though, there's a crapload of changes that would need be done to enforce this example above)
< gwillen>
*nods*
< sipa>
but adding const in places where the code already works has no effect except add assurances
< tryphe>
alright, that makes sense. i guess there's enough tests anyways to catch most issues.
< gmaxwell>
you can also introduce bugs by peppering around const, so it's not like those changes are free to review, even ignoring the effect on change flow.
< sipa>
in general adding const can't have too much effect if the code works both before and after... but there are some
< sipa>
for example moving a const argument will silently introduce a (deep) copy of certain objects, where a move was expected
< tryphe>
yeah, the problem is knowing the intent of the code you are modifying
< sipa>
tryphe: exactly!
< sipa>
adding such things where they are intended is great, but adding them mechanically is mostly useless without careful review of every individual case
< gmaxwell>
That is also a general problem with broad 'cleanup' changes, they're done without much understanding or care about the code they're changing.
< tryphe>
one good reason i like to do it, though, is because i get a compiler error when i try to modify something that wasn't intended by the programmer. it's easy for someone to... misuse the code i guess is a better way to put it.
< tryphe>
sorta like using nullptr instead of 0, you state the intention of the usage for future programmers to reference, instead of relying on them to make a good guess, hehe
< tryphe>
good/correct*
< bitcoin-git>
[bitcoin] Bushstar opened pull request #15979: Replace incorrectly named comment with func (master...replace-with-func) https://github.com/bitcoin/bitcoin/pull/15979
< instagibbs>
provoostenator, meshcollider re-review beg of #15452
< gribble>
https://github.com/bitcoin/bitcoin/issues/15452 | Replace CScriptID and CKeyID in CTxDestination with dedicated types by instagibbs · Pull Request #15452 · bitcoin/bitcoin · GitHub
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #15982: test: Remove True argument to CBlock::serialize (master...1905-testWitnessTrue) https://github.com/bitcoin/bitcoin/pull/15982
< luke-jr>
does C++ have a way to say "this method can be called with a const or non-const `this`, but the types yielded by it have the same constness"?
< jb55>
can you re-org from genesis in principle? there's nothing preventing that in the code right?
< sipa>
genesis is hardcoded, you can't reorg away from the genesis block
< sipa>
even if you disable checkpoints
< jb55>
I mean from one after genesis
< sipa>
yes
< jb55>
kk thx
< luke-jr>
jb55: checkpoints prevent that, but can be disabled as sipa points out
< phantomcircuit>
luke-jr, iirc the checkpoints don't prevent a re-org like that either exactly
< sipa>
they do
< sipa>
assumevalid does not
< sipa>
but checkpoints actually prevent reorging away feom
< MarcoFalke>
gmaxwell: Wouldn't the (blocksonly) border node need to accept txs from whitelisted wallets?
< MarcoFalke>
Also, if blocksonly is recommended for security or resource conversation, it shouldn't be a hidden debug option
< MarcoFalke>
Anyway, I tend to agree with your idea to make whitelisting more specific
< MarcoFalke>
But that should probably happen on a per-node basis. E.g. whitelistfortxrelay=${Node_Address}:{Whitelist_Level}
< gmaxwell>
It shouldn't be hidden.
< gmaxwell>
MarcoFalke: not levels, flags, most things whitelisty things are not ordereable.
< gmaxwell>
It's hidden because of historical reasons, I'd forgotten it was. it's frequently recommended.
< MarcoFalke>
So that should be changed first, I guess
< gmaxwell>
Flags would be pretty awesome.
< bitcoin-git>
[bitcoin] sipa opened pull request #15986: Add unmodified-descriptor-with-checksum to getdescriptorinfo (master...201905_justchecksum) https://github.com/bitcoin/bitcoin/pull/15986
< bitcoin-git>
[bitcoin] luke-jr opened pull request #15987: Wallet, GUI: Warn when sending to already-used Bitcoin addresses (also RPC: include such information in getaddressinfo) (master...wallet_no_reuse) https://github.com/bitcoin/bitcoin/pull/15987