< fanquake>
promag restarted all the tests in 15153
< fanquake>
planning on testing that again
< promag>
fanquake: I'll push shortly
< promag>
I think it's good enough for 1st GUI support
< promag>
in order to support "cancel open because it's rescanning" we have to do some refactors
< fanquake>
promag, yes I think so. There are still edge cases/improvements that can be fixed/made later
< fanquake>
i.e closing a modal after opening a wallet, then opening the same wallet again
< fanquake>
You end up with the first completing, then the second opening failing because the same wallet is loaded already.
< promag>
fanquake: I think I can fix that, by disabling the close
< promag>
fanquake: right
< fanquake>
Disabling the close button could work, although that's probably not ideal if you are stuck with a long time to rescan wallet, and want to bail out.
< promag>
fanquake: ryanofsky: I think Node::loadWallet should return the Wallet immediately, and the the GUI could add handlers and interact with the wallet asynchronously
< promag>
fanquake: ah I see, then I it should not be modal, and if the same wallet is opened then it fails
< wumpus>
hhmm translation process for adding new translations needs to be updated for the MSVC build system
< wumpus>
translations are listed in three places now :(
< wumpus>
(working on this)
< bitcoin-git>
[bitcoin] benthecarman opened pull request #15337: rpc: Fix for segfault if combinepsbt called with empty inputs (master...rpc_combinepbst_empty_fix) https://github.com/bitcoin/bitcoin/pull/15337
< luke-jr>
dongcarl: it shouldn't affect build times *too* much, so long as it uses the same build dir between commits
< dongcarl>
Okay, I will PR to do `git --exec` then
< dongcarl>
or rather `git rebase --exec`
< instagibbs>
I don't have a lot of experience in this but how often has this been a problem
< dongcarl>
instagibbs: You mean git bisectability?
< luke-jr>
that's a good point too
< wumpus>
I don't see a strong reason to do this either
< instagibbs>
dongcarl, correct, how often has this been an issue
< luke-jr>
AFAIK git bisect doesn't go inside merges unless the issue is part of that set of commits
< wumpus>
testing the top commit of a PR should be enough, anything else is for certain going to confuse people, how many projects run the CI on every single commit even in merges?
< dongcarl>
I guess I was thinking that if we had a long-standing bug we could use bisect to find where it started
< luke-jr>
too bad there's no way to set priorities so Travis just runs everything on intermediate commits if it gets idle cycles
< luke-jr>
dongcarl: you can do that regardless
< wumpus>
bisect is already quite usable
< wumpus>
never had problems really
< instagibbs>
dongcarl, I believe luke is right in that it can still do it, regardless
< dongcarl>
Okay cool
< instagibbs>
as long as merge commits are properly passing
< dongcarl>
Can anyone tell me what the deprecated variable naming convention is? I see that `v*` is vector... `f*` seems to be booleans? There's also a `vf` in net.cpp, which is an array of booleans?
< dongcarl>
I'm just reading thru net.cpp adding comments
< sipa>
there are many variables whose type changed without the name changing accordingly
< sipa>
there's probably some fun archaeology to do with that
< dongcarl>
"fun"
< wumpus>
dongcarl: btw if you intend to submit more comment changes to net code, that'd be very helpful, but it's probably better to group them into one PR instead of opening separate ones like #15343
< promag>
dongcarl: apologies for that, you were much happier before
< sipa>
dongcarl: so psz is pointer-to-zero-terminated-string
< sipa>
which used to occur frequently
< dongcarl>
sipa: My mom warned me about nasty C-style zero-terminated strings! They hang out in all the bad stacks!
< wumpus>
hebasto: *conceptually* it's a good idea, it's just virtually impossible to implement correctly with gettext
< hebasto>
wumpus: thanks
< wumpus>
we've run into a similar issue in the past for the status bar, where we wanted to return the untranslated error on the RPC and the translated one on the GUI
< wumpus>
ended up duplicating the text ...
< wumpus>
would certainly *not* be nice to do for all errors