Re: [Development] Builds failing due to android host

[Roscher-Nielsen Nils Christian]

Hi,

Do we have something like a production and a test environment, so that we can 
push changes, updates, upgrades etc. to the test environment to see if it 
works, before using it in production? Realistically things will fail more often 
than we hope, and as Tony say, there is usually something waiting in the 
queue.. Or do we not have enough HW etc. to realistically do this?

I know we do this for other of our internal tools, but they might be less 
"complicated". Is it an idea worth investigating (when we have the time :) )

Best regards
Nils Christian Roscher-Nielsen

Sales Engineer - Digia, Qt

Visit us on: http://qt.digia.com

From: development-bounces+nils.roscher-nielsen=digia@qt-project.org 
[mailto:development-bounces+nils.roscher-nielsen=digia@qt-project.org] On 
Behalf Of Shaw Andy
Sent: 22. mai 2013 11:54
To: Sarajärvi Tony; development@qt-project.org
Subject: Re: [Development] Builds failing due to android host

I grant you that we are always releasing something, but in this instance I was 
referring to the fact that 5.1 beta is already out, with the final due in just 
a few weeks and time is running short because of the timing of that then it 
might be best to hold off on doing some upgrades until a generally more safer 
time. As such now the merge from stable to release as to wait because of the CI 
trouble, which puts more pressure on the actual schedule.

Andy

From: Sarajärvi Tony
Sent: 22. mai 2013 11:45
To: Shaw Andy; development@qt-project.org
Subject: RE: Builds failing due to android host

Do we ever have a situation that something is not "late into a release 
process"? If it's not 5.1, it's 5.1.1, or 5.2 beta, or 5.2 RC, or 5.2.1...  not 
to mention our 4.x builds :)

-Tony

From: Shaw Andy
Sent: 22. toukokuuta 2013 12:42
To: Sarajärvi Tony; development@qt-project.org
Subject: RE: Builds failing due to android host

That doesn't make it a valid thing to still do, testing or not. I am not sure 
what others think but do we want to still do upgrades to things on CI this late 
into a release process?

Andy

From: Sarajärvi Tony
Sent: 22. mai 2013 10:54
To: Shaw Andy; development@qt-project.org
Subject: RE: Builds failing due to android host

We have always something coming ;) And we didn't realize that this could cause 
problems. We got a request to update the tool chain and we just did as 
requested. We should have tested this on the dev side, but we don't have the 
luxury of time on our side. From now on, we don't upgrade or update anything 
before verifying on the dev side.

-Tony

From: Shaw Andy
Sent: 22. toukokuuta 2013 11:32
To: Sarajärvi Tony; development@qt-project.org
Subject: RE: Builds failing due to android host

Maybe I missed something, but why are upgrades like this being done in the 
middle of a release cycle? With Qt 5.1 final due in not too long then shouldn't 
these kind of upgrades wait until after the final release in order to avoid 
this sort of thing being a problem for the release?

Andy

From: development-bounces+andy.shaw=digia@qt-project.org 
[mailto:development-bounces+andy.shaw=digia@qt-project.org] On Behalf Of 
Sarajärvi Tony
Sent: 22. mai 2013 10:27
To: development@qt-project.org
Subject: Re: [Development] Builds failing due to android host

Updating the NDK began causing new and new problems, so Android builds are 
marked as 'forcesuccess' for the time being.

Regards,
-Tony

From: Sarajärvi Tony
Sent: 21. toukokuuta 2013 13:07
To: development@qt-project.org
Subject: Builds failing due to android host

Hi

If your build failed due to this message "Can not detect the android host. 
Please use -android-ndk-host option to specify one" you can just safely restage 
it. We made changes to our Puppet configuration regarding Androids, and yes, 
the CI team caused this fail.

We have a small known problem with Puppet, that currently running builds don't 
prevent Puppet from running.
Due to this, Puppet could potentially remove compile tools from under the 
build. Also, when a build finishes, it reboots the machine. This reboot kills 
the Puppet process and everything it was doing currently. This is most likely 
what happened to Android tools. They were being upgraded to 64 bit tools, and 
on some machines Puppet was killed during this process.

It will automatically retry in 90 minutes (which is Puppet's current timeout). 
Thus at some point it will fix itself.

We have a change for this already inbound so that Puppet doesn't do this, but 
it's been incoming since February, so don't hold your breath ;)

-Tony




Tony Sarajärvi
Senior Software Designer
Digia, Qt

Digia Plc
Elektroniikkatie 10
FI-90590 Oulu

Email: tony.saraja...@digia.com
Mobile: +358 050 482 1416
http://qt.digia.com
| Blog | Twitter | 
LinkedIn | 
YouTube | 
Facebook

Re: [Development] QtMultimedia BIC / header cleanliness issue

[Roscher-Nielsen Nils Christian]

> -Original Message-
> 
> > On Tuesday, January 22, 2013 11:40:23 AM Oswald Buddenhagen wrote:
> > > On Mon, Jan 21, 2013 at 08:14:03AM -0800, Thiago Macieira wrote:
> > > > On segunda-feira, 21 de janeiro de 2013 15.33.59, Knoll Lars wrote:
> > > > > Finally reading up on some old emails…
> > > > >
> > > > > I'd say we add the virtual destructors. Better to deal with the 
> > > > > fallout
> > > > > now then in the future.
> > > >
> > > > And rename the libraries to libQt5Multimedia.so.6 ?
> > >
> > > imo no. just pretend that it never happened.
> > > at this point there aren't many packages which are considered stable,
> > > and even fewer packages which depend on this module. other users
> won't
> > > be affected by this change to any noteworthy degree.
> > > however, make sure this gets into 5.0.1, which means bugging the
> release
> > > team at this point, very soon.
> >
> > That seems most reasonable to me. It's not perfect-by-the-book, but I also
> > don't think anyone would notice it given the release just in december.
> > (famous
> > last words!)
> 
> I don't think it should be done silently at all, I'm not exactly for the idea 
> of
> even breaking BC but I can accept an exception this early in the Qt 5.x series
> if there is no other way and it is critical to be done.  But this should be 
> clear in
> the changelog and release notes that it was done, better to be open about it
> than to hope no-one notices.

I must say I agree with Andy here. Especially as we are actually doing this 
development in the open. As anyone can watch us, we are setting a standard for 
how SW development is done, and Qt has had a high star as far as SW quality 
goes. This is the reason many people choose Qt. 

And I think that is goes without saying, even though not many people are likely 
to be using it, that especially as we do something out of the ordinary we need 
to properly document it. Anything else would be rude. 

Best regards
 Nils Christian Roscher-Nielsen

  – Digia, Qt


> ___
> Development mailing list
> Development@qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development
___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development

[Development] Building PostgreSQL plugin by default

[Roscher-Nielsen Nils Christian]

Hi all,

Can we start building the PostgreSQL plugin in our binary packages again by 
default?
We used to do this before, but stopped doing it at some point. Was there a good 
reason for that, and is it still relevant?

Thanks.

Best regards
Nils Christian Roscher-Nielsen

Sales Engineer - Digia, Qt

___
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development