< bitcoin-git>
[bitcoin] sipa opened pull request #18861: Do not answer GETDATA for to-be-announced tx (master...202004_private_getdata) https://github.com/bitcoin/bitcoin/pull/18861
< bitcoin-git>
[bitcoin] fanquake opened pull request #18862: Remove fdelt_chk back-compat code and sanity check (master...remove_fdelt_chk_back_compat) https://github.com/bitcoin/bitcoin/pull/18862
< bitcoin-git>
[bitcoin] promag opened pull request #18863: refactor: Drop unused CScriptVisitor return type (master...2020-04-cscript-visitor) https://github.com/bitcoin/bitcoin/pull/18863
< elichai2>
Is practicalswift on IRC?
< elichai2>
I'm trying to figure out some fuzzing best practices, is it ok to reuse the same fuzzing input on multiple functions or should I "pull" more data for each function I call? (cc BlueMatt ) should I minimize the amount of fuzzing input I'm using or maximize it?
< jonasschnelli>
elichai2: he is sometimes. But no bouncer I guess.
< bitcoin-git>
[bitcoin] Sjors opened pull request #18864: Add v0.16.3 backwards compatibility test, bump v0.19.0.1 to v0.19.1 (master...2020/05/backwards_compat) https://github.com/bitcoin/bitcoin/pull/18864
< wumpus>
elichai2: correct, const is always advisory in c++ anyway
< elichai2>
wumpus: you mean as function input, not variable declaration, right?
< elichai2>
(or as pointer qualifier or however you want to call that hehe)
< wumpus>
I mean in general, you can always cast it away (of course it's a bad idea)
< bitcoin-git>
[bitcoin] MarcoFalke closed pull request #18863: refactor: Drop unused CScriptVisitor return type (master...2020-04-cscript-visitor) https://github.com/bitcoin/bitcoin/pull/18863
< bitcoin-git>
[bitcoin] MarcoFalke reopened pull request #18863: refactor: Drop unused CScriptVisitor return type (master...2020-04-cscript-visitor) https://github.com/bitcoin/bitcoin/pull/18863
< luke-jr>
wumpus: I'm not sure you are guaranteed that?
< luke-jr>
IIRC string literals at least can't just be cast away constness
< BlueMatt>
elichai2: its mostly a chance-of-crash per input bit thing - the fuzzer can only try X input bits per time, so if you get more of a chance of a crash from being less verbose, do it (of course it also depends on the type of input - i have a bunch of fuzz targets that don't do well with fuzzer's traditional input bit permutations that focus on arithmetic etc)
< wumpus>
did we change anything to linking of the tests recently? re #18869
< wumpus>
luke-jr: my point wasn't that it would be safe to do so
< sipa>
wumpus: const is not always advisory in C++
< wumpus>
okay
< sipa>
if you have a const-declared variable (not just a const reference/pointer to it), then modifying it is UB
< wumpus>
I wasn't planning to do anything like that
< sipa>
i believe the spec says that modifying through a const pointer/reference is legal only if you know something else in the program holds a non-const reference to the variable
< wumpus>
I'm aware the compiler can make all kinds of optimizations based on const
< sipa>
usually it can't, actually
< luke-jr>
sipa: interesting, so a compiler that tagged const references and aborted if you tried to remove the constness, would be non-compliant?
< sipa>
luke-jr: removing the constness is always fine; modifying through a const-reference casted to non-const is UB if no other non-const references exist
< sipa>
but say you have a struct X { const int c; X(int v) : c(v) {} };... then modifying c would always be UB
< wumpus>
so does anyone have an idea about the link issue?
< jeremyrubin>
I think it looks like maybe there is disagreement on utility in core but maybe could be something to build for knots and then roll into core if it works well
< bitcoin-git>
[bitcoin] achow101 opened pull request #18870: build: Allow BDB between 4.7 and 5.3 without --with-incompatible-bdb (master...allow-other-bdb) https://github.com/bitcoin/bitcoin/pull/18870
< luke-jr>
jeremyrubin: I tend to agree it sounds like overkill - but if you want to maintain it for Knots, I guess there's no hurt experimenting
< bitcoin-git>
[bitcoin] brakmic opened pull request #18871: rpc: prevent circular deps by extracting into separate include (master...prevent-circular-deps) https://github.com/bitcoin/bitcoin/pull/18871