< andytoshi> sipa: ahh ok, thanks
< meshcollider> why is the boost::to_lower() inside a #ifdef WIN32 here: https://github.com/bitcoin/bitcoin/blob/master/src/util.cpp#L419
< meshcollider> does that mean its only case insensitive on win32
< Lightsword> windows itself is mostly case insensitive
< Lightsword> although I think some of the underlying filesystems used like NTFS can be case sensivite
< sipa> NTFS on windows is case insensitive but case preserving
< sipa> but that line of code has nothing do with filesystems
< sipa> it's so that "bitcoind /daemon=1" works, which I think gets passed to the application as "/DAEMON=1"
< meshcollider> ah I see
< bitcoin-git> [bitcoin] irq13 opened pull request #11860: move banned peer log messages behind -debug=net (master...no-log-bans) https://github.com/bitcoin/bitcoin/pull/11860
< ieririr> hi there, does anybody have an idea of the time needed to spread a fix? in case of major flaw in crypto core I mean
< promag> is it me or funding an already funded transaction hits AvailableCoins?
< andytoshi> sipa: /win 23
< andytoshi> oops sorry
< BGL> is it by design that the core client gives no indication that it's creating a new wallet file (ie, when one goes missing) ? or is that a bug?
< sipa> BGL: what would you expect to happen?
< BGL> at a minimum say hey your wallet file was missing?
< sipa> i think it does say that in the log, but maybe not
< BGL> certain things have a real minimum-effort feel going for it with the client
< BGL> and with btc being 15k now i think the standards might need to be up a little
< sipa> what else would you expect?
< BGL> so like let's say your sql db goes down
< BGL> <insert variable of obvious common sense>
< BGL> asking what people expect is not lending toward resolution
< sipa> i don't understand what you feel is missing
< BGL> that may be exactly the problem
< sipa> okay
< sipa> well, issue reports or pull requests welcome
< Randolf> BGL: Are there some specific problems with the standards that you are aware of presently?
< BGL> waiting on people to point out the obvious is often the demise of software
< sipa> what do you expect to happen? fail to start?
< sipa> that would break backward compatibility
< BGL> yes, let's all act as if your wallet file disappearing and the client literally indicating nothing is wrong is perfectly acceptable.
< BGL> i would literally fire both of you for acting so stupid
< sipa> the client is unable to distinguish that from starting up the first time
< BGL> the correct answer is yes, maybe that's a good idea, we should look into that
< BGL> maybe that's part of the problem?
< BGL> does this sound like such a difficult issue?
< sipa> yes
< BGL> wow
< sipa> there is literally nothing the client could do the know that missing a file is different from starting up the first time
< Randolf> BGL: I'm not sure who you're directing that to, but please keep in mind that people here are volunteers.
< sipa> you're very welcome to help improve the software
< sipa> but insulting volunteers is not likely to get your issue resolved
< BGL> being a volunteer does not make up for incompetent ego
< BGL> my only issue is the quality of this client
< Randolf> BGL: With regard to people losing their wallet files, backups are highly recommended, and having copies of one's Private Keys is obviously imperative. Perhaps if this isn't being explained to the end-user properly there could indeed be room for improvement there.
< sipa> BGL: you must literally have manually deleted a database file
< sipa> there is no other way the file would be missing
< sipa> that is the exact same situation as starting up the first time
< sipa> please, be constructive
< Randolf> BGL: My ego doesn't factor into this -- it went on vacation a long time ago, and I have no idea where it is these days, although I do keep receiving hotel bills so I know where it has been. ;)
< Randolf> sipa: There is another possibility -- the user logged in with a different profile and so it looks like a new setup to the client software.
< gmaxwell> I've never previously heard a report from a user that their wallet file was unexpectedly removed. Any idea how a user would end up in that situation?
< Randolf> BGL: Did your wallet file go missing?
< sipa> Randolf: "profile" ?
< Randolf> sipa: Well, in the Windows world at least, there are user profiles which each have their own collections of home directories and personalized settings for the OS and various applications.
< Randolf> sipa: Unfortunately with Windows, whenever it corrupts itself the user is sometimes logged in with a temporary profile instead of their usual one. In such cases, a wallet file could seem to go missing if it was stored under the user-specific home directory.
< Randolf> sipa: But this really isn't a Bitcoin issue -- it's an Operating System issue that has the same impact on other applications too.
< gmaxwell> Randolf: how would that deny access to the wallet but also leave the blockchain?
< Randolf> gmaxwell: That's interesting. If the blockchain was also stored in the user's profile directory, then your concern makes sense. However, if the blockchain was stored in a public profile (as some applications with shared or constant data do) then that could explain why the blockchain would remain
< Randolf> in place.
< Randolf> gmaxwell: Another possibility is that bitcoind has its own profile and is storing the blockchain data under its own profile which is not the same as the user's profile.
< Randolf> ...specifically the wallet-user's profile.
< sipa> Randolf: unless he's running unreleased software, the blockchain and the wallet are in the same directory
< Randolf> sipa: Okay, then that rules that possibility out.
< Randolf> Another possibility is that the user installed some malicious software that uploads wallet files then deletes them from the user's system.
< BGL> Randolf: that's not possible because <the user> "must literally have manually deleted a database file"
< sipa> BGL: if that's not what happened, i'd very much like to know
< BGL> please save me from having to explain an infinite number of ways files can be deleted
< sipa> not by bitcoind itself
< sipa> either you directly, or software you chose to run deleted a file
< BGL> reminder: i'm simply pointing out the client says/does nothing when a wallet file goes missing
< sipa> yes, because *it can't*
< mesh_> Best we could do is add a debug.log message which says something like "wallet.dat does not exist, creating..."
< mesh_> BGL: that could be an opportunity for you to contribute, you could open a PR if you wanted :)
< gmaxwell> mesh_: the logs do reflect when it creates one.
< gmaxwell> The software also does not warn you when your computer is running in an enviroment without enough oxygen for human life... there are an infinitude of ways that something strange could be going on; without an understood sequence of events that would cause the situation it's usually not reasonable to do anything about it, since the "do something" is usually context dependant.
< mryandao> ^ lol.
< meshcollider> gmaxwell: I can only see a message like "2017-12-10 22:16:37 Using wallet test.dat" and then it reports there are 0 keys, but no actual creation message?
< gmaxwell> Morover, spurrious warnings cause user error. E.g. we've seen users lose funds because they opened their debug.log and saw log entries with "ERROR" in them, and then started taking increasingly drastic actions to remediate the actually-harmless messages which eventually culminated in them forever deleting their wallet, or posting their files online for othe people ... including the wallet.
< gmaxwell> Moreover*
< gmaxwell> meshcollider: indeed, it shows it with zero sizes when its first created.
< mryandao> maybe bitcoin-core should build in telemetry service and leave the generation of debug log for advanced users.
< meshcollider> gmaxwell: yeah but that is not clear to any user, I'm just suggesting BGL might want to add a simpler message
< mryandao> to protect users against themselves
< gmaxwell> mryandao: well we intentionally toned down the text in debug log to reduce incidents of those sorts of screwups, seems to have owrked.
< gmaxwell> meshcollider: would be reasonable and harmless, though I'm dubious that it would actually help anything.
< Randolf> BGL: Doesn't the client report to you that the file was not found?
< Randolf> BGL: ...when you attempt to open it?
< sipa> Randolf: there is no 'opening' the file
< sipa> there is a hardcoded filename (wallet.dat) which contains the wallet; if it doesn't exist, it's created
< Randolf> sipa: Ah, well, in that case the client software won't be able to detect if a previous wallet file was deleted.
< sipa> the daemon supports having multiple wallets loaded simultaneously, but the GUI doesn't support opening/closing other wallets
< sipa> (not yet, at least)
< Randolf> Okay.
< Randolf> That's very curious. I wonder what happened to BGL's wallet file. I'm sure that neither the client nor bitcoind deleted it.
< Randolf> BGL: Did you save your Private Keys?
< Randolf> BGL: If you did, then you can enter them again.