00:05 < emilengler> IMO we should add a fat red banner to tell the users that they should really verify there downloads

00:05 < harding> wumpus: FYI, release PR for BitcoinCore.org: https://github.com/bitcoin-core/bitcoincore.org/pull/679

01:10 < bitcoin-git> [bitcoin] RandyMcMillan opened pull request #17424: doc:Travis CI note on Build Timeouts (master...travis-note) https://github.com/bitcoin/bitcoin/pull/17424

11:06 < bitcoin-git> [bitcoin] hebasto opened pull request #17427: qt: Fix missing qRegisterMetaType for size_t (master...20191109-fix-signal-argument-type) https://github.com/bitcoin/bitcoin/pull/17427

11:25 < JeremyCrookshank> Hello

16:46 < fanquake> I wonder if anyone testing the GUI during the 0.19.0 rcs opened on the debug window

16:47 < fanquake> Something in there has been broken for ~4 months now

16:50 < fanquake> I think we definitely put too much weight on the fact that so few bugs/issues are reported during release candidates. I feel like it means more that no-one is actually doing any testing rather than we are releasing non-buggy software.

16:51 < fanquake> Granted, problems with the gui aren't exactly a high priority.

16:51 < sipa> what is broken about it?

16:52 < fanquake> The mempool stats have been showing as "N/A"

16:53 < fanquake> Not exactly critical, but also the kind of thing you'd assume would be noticed if anyone had opened the window

16:53 < fanquake> Looks like the signal that updates the gui was broken in #16348

16:53 < gribble> https://github.com/bitcoin/bitcoin/issues/16348 | qt: Assert QMetaObject::invokeMethod result by promag . Pull Request #16348 . bitcoin/bitcoin . GitHub

16:54 < fanquake> i.e: https://imgur.com/a/u0UyKUx

17:15 < promag> fanquake: why that PR broke?

17:15 < fanquake> promag: see #17427

17:15 < gribble> https://github.com/bitcoin/bitcoin/issues/17427 | qt: Fix missing qRegisterMetaType for size_t by hebasto . Pull Request #17427 . bitcoin/bitcoin . GitHub

17:16 < promag> still, #16348 only adds some asserts and qRegisterMetaType

17:16 < gribble> https://github.com/bitcoin/bitcoin/issues/16348 | qt: Assert QMetaObject::invokeMethod result by promag . Pull Request #16348 . bitcoin/bitcoin . GitHub

17:18 < fanquake> yea, maybe it's been broken for longer than that then. Hard to keep up with the constant gui refectoring

17:18 < promag> I don't think so

17:18 < promag> I think it was caused by #17135

17:18 < gribble> https://github.com/bitcoin/bitcoin/issues/17135 | gui: Make polling in ClientModel asynchronous by promag . Pull Request #17135 . bitcoin/bitcoin . GitHub

17:19 < promag> because now the signal mempoolSizeChanged is enqueued in the gui event loop

17:19 < promag> and that was failing

17:20 < fanquake> and the fix is registering a new type def ?

17:20 < promag> I can't check atm, but I bet before 17135 it works

17:20 < fanquake> I'll take a look

17:21 < promag> from https://doc.qt.io/qt-5/qmetatype.html#qRegisterMetaType

17:21 < promag> This function is useful to register typedefs so they can be used by QMetaProperty, or in QueuedConnections

17:21 < promag> voila

17:22 < fanquake> heh, glad we managed to sneak that into 0.19 too

17:22 < fanquake> or backport it in actually

17:23 < promag> Not sure if possible but I think QT_FATAL_WARNINGS=1 should be always set

17:24 < promag> I don't see any downside of that

17:25 < fanquake> could probably add a note to the dev docs for anyone doing gui work

17:26 < promag> fanquake: I mean hardcoded

17:28 < fanquake> promag: In release binaries? I'm not sure that's a good idea unless you can guarantee it'll never crash. Otherwise Qt could just through some dumb warning about missing fonts etc, and crash bitcoin-qt

17:29 < promag> Really? If we require a font then the font is also shipped/embeded right?

17:30 < fanquake> I don't think so

17:31 < fanquake> We ask the OS for certain font types when we need them

17:33 < fanquake> i.e in GUIUtil::fixedPitchFont().

17:34 < promag> ok

17:38 < fanquake> Looks like the worse case, if one of those calls was to fail, is Qt just picks/falls back to some other trash looking font

17:39 < fanquake> https://code.qt.io/cgit/qt/qtbase.git/tree/src/gui/text/qfontdatabase.cpp#n2548

18:31 < luke-jr> promag: end users should never be subject to warnings crashing..

18:31 < luke-jr> pretty sure bitcoin-qt always has warnings for me, too

18:32 < luke-jr> only reason I could see to embed a font is for tonal support, and Core doesn't support that (Knots does, but still doesn't bundle - it just turns off tonal support if no font is found)

22:46 < bitcoin-git> [bitcoin] hebasto opened pull request #17428: p2p: Try to preserve outbound block-relay-only connections during restart (master...20191109-anchors) https://github.com/bitcoin/bitcoin/pull/17428