Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 09:08:50 PDT Thiago Macieira wrote: > And maybe a fix for MSVC's inability to override the global operator new. > See https://bugreports.qt.io/browse/QTBUG-63274. Nope, that doesn't work either. Proof: https://godbolt.org/g/FM9ohw Note how it calls the DLL:

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 09:08:50 PDT Thiago Macieira wrote: > On Tuesday, 19 September 2017 08:21:28 PDT Oswald Buddenhagen wrote: > > On Tue, Sep 19, 2017 at 07:49:58AM -0700, Thiago Macieira wrote: > > > Q_ALWAYS_INLINE, however, has no bearing in ABI. It's a purely > > > optimisation > >

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 08:21:28 PDT Oswald Buddenhagen wrote: > On Tue, Sep 19, 2017 at 07:49:58AM -0700, Thiago Macieira wrote: > > Q_ALWAYS_INLINE, however, has no bearing in ABI. It's a purely > > optimisation > > feature and can be added or removed at will. > > would you bet on that?

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 08:20:50 PDT Edward Welbourne wrote: > > Q_REQUIRED_RESULT being moved to the beginning of the line was a very > > restricted change. There was exactly one commit affecting about 20 > > functions, that's all. > > All the same, if it gets added to a function (rather

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Oswald Buddenhagen
On Tue, Sep 19, 2017 at 07:49:58AM -0700, Thiago Macieira wrote: > Q_ALWAYS_INLINE, however, has no bearing in ABI. It's a purely optimisation > feature and can be added or removed at will. > would you bet on that? are you _sure_ it does not affect the emission of out-of-line copies, on any

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Edward Welbourne
Thiago Macieira (19 September 2017 16:49) > The problem with Q_DECL_NOTHROW is that it once added, it cannot be > removed. There's also the choice: we choose to add it only to > functions that (in addition to neevr throwing) are wide contract. So > we should see it and review that kind of change.

Re: [Development] QRandomGenerator and boot times

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 03:16:32 PDT Konstantin Tokarev wrote: > > Why? Is hash randomization really important in your case? > > Actually, you can read value of QT_HASH_SEED from /dev/urandom, restoring > previous behavior exactly. Read what I wrote in my emails and in the Python bug

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Thiago Macieira
On Tuesday, 19 September 2017 03:02:16 PDT Edward Welbourne wrote: > OK, but I infer there's no harm in including them in the list of tokens > the script already ignores - explicit, Q_ALWAYS_INLINE, Q_DECL_NOTHROW > before the new additions. How about Q_REQUIRED_RESULT ? The problem with

Re: [Development] Coding style for lambdas, particularly their open-brace

2017-09-19 Thread Edward Welbourne
On Fri, Aug 18, 2017 at 2:17 PM, Edward Welbourne > wrote: >> We have a draft policy for lambdas at [0], in a section that begins with >> >> Note: This section is not an accepted convention yet. >> This section serves as baseline for

Re: [Development] QRandomGenerator and boot times

2017-09-19 Thread Sami Nurmenniemi
On 19.09.2017 13:16, Konstantin Tokarev wrote: 19.09.2017, 13:14, "Konstantin Tokarev" : 14.09.2017, 10:22, "Sami Nurmenniemi" :  Hi,  Commit 120ecc976fc3d5504d234702f68c2ad3898b77a4 changes default behavior of QRandomGenerator to use getentropy

Re: [Development] QRandomGenerator and boot times

2017-09-19 Thread Konstantin Tokarev
19.09.2017, 13:14, "Konstantin Tokarev" : > 14.09.2017, 10:22, "Sami Nurmenniemi" : >>  Hi, >> >>  Commit 120ecc976fc3d5504d234702f68c2ad3898b77a4 changes default behavior of >> QRandomGenerator to use getentropy instead of /dev/urandom. This causes

Re: [Development] QRandomGenerator and boot times

2017-09-19 Thread Konstantin Tokarev
14.09.2017, 10:22, "Sami Nurmenniemi" : > Hi, > > Commit 120ecc976fc3d5504d234702f68c2ad3898b77a4 changes default behavior of > QRandomGenerator to use getentropy instead of /dev/urandom. This causes > problems for device boot times when using QRandomGenerator in the

Re: [Development] Meeting minutes from Qt Release Team meeting 05.09.2017

2017-09-19 Thread Edward Welbourne
On quarta-feira, 13 de setembro de 2017 01:54:12 PDT Edward Welbourne wrote: >> [...] Should Q_DECL_CONST_FUNCTION >> changes be ignored ? Q_NORETURN ? They currently aren't. Thiago Macieira (13 September 2017 15:51) > Those two are optimisations and could safely be ignored, but they > don't

[Development] Qt 5.6.3 packages available

2017-09-19 Thread Jani Heikkinen
Hi all, We have now Qt 5.6.3 packages available via online installer ( under 'preview' node still) Target is to get Qt 5.6.3 release out still during this week so please inform us immediately if there seems to be something new broken in these packages. And please fill Qt 5.6.3 known issues