[Interest] Let's talk about the upcoming Camera breakage in 5.4.2

Jason H jhihn at gmx.com
Wed May 13 16:54:20 CEST 2015


I can appreciate that. I do. However, I think most of the people tracking the x.x.Y releases are likely to be mobile people, where there are a lot of mobile-specific fixes going on, including lot of image capturing on those mobile platforms. The Desktop is of course capable of capturing images as well (I use As the 5.4.1 code "fixed" the camera capture issue, desktop to build the app). But those of us doing mobile stuff very often need the latest. I'd expect everyone who now has that code base to be dependent on it. But we also need other fixes in 5.4.2. 

I suspect that the impact of reverting correct capture in 5.4.1 is greater than those waiting for it to be broken again 5.4.2. Which is why I raised the issue on this list. How many people are skipping/have workarounds for 5.4.1, whereas everyone doing capture properly on 5.4.1 will now be broken?

I think we need to determine and proceed the path of least impact. If I'm wrong, I'll back off. But since there are so many mobile fixes in the x.x.Y releases, I have a hard time believing that is the case.

> Sent: Wednesday, May 13, 2015 at 8:55 AM
> From: "Gunnar Sletta" <gunnar at sletta.org>
> To: "Jason H" <jhihn at gmx.com>
> Cc: "Interests Qt" <interest at qt-project.org>
> Subject: Re: [Interest] Let's talk about the upcoming Camera breakage in 5.4.2
>
> I’m sorry you are hit by this, but the idea was that people coming from 5.3 to 5.4 would not suffer regressions and neither people moving from 4.8, 5.0, 5.1, 5.2 and 5.3 -> 5.5 or from 5.4 LTS to 5.5 or 5.6 or any other 5.6+ which already had workarounds in place which would have now been broken. It is all about not breaking existing code. Yes, there will be a few people who are affected by the fact that there was an well-intended incompatibility introduced to 5.4.1 in the first place, but this only affects those that are tracking the latest patch releases, and those are not in majority.
> 
> cheers,
> Gunnar
> 
> > On 12 May 2015, at 17:42, Jason H <jhihn at gmx.com> wrote:
> > 
> > I wanted to open this discussion up to anyone who is using a Camera, who has working code in 5.4.1.
> > 
> > Brief backstory:
> > At 5.4.1 introduced a fix that is is being removed in 5.4.2, because it is considered a "behavior" change and not a bug fix. I think this categorization is wrong, and I'm going to explain why.
> > The change pertains to the EXIF rotation header. So if you capture in portraid mode, or upside down, Qt is aware of it. In 5.4.2, this is being removed.
> > 
> > The removal of this orientation flag mean now that images won't be captured correctly, as they will have the incorrect aspect ratio from what you expect. This has a rippling effect in that no longer can you capture a picture pf the preview and display it pixel-for-pixel on a subsequent screen. It is not just a rotation issue. I've done the rotation fix (which is quite simple) and it messes more things up than that. This causes additional code to be written by Qt customers to target 5.4.2 specifically.
> > 
> > But here's the kicker. The orientation is being put back in to 5.5. Meaning that customers who were looking for a few bug fixes in 5.4.2, will have to write additional code for one version of Qt, then back it out for 5.5. I think for all the customers doing camera capture on 5.4.1 (I assume this is all of the customers using camera capture) we will need to write code on 5.4.2, then back it out. 
> > 
> > Ideally, I'd like to get the EXIF orientation put back into in 5.4.2.
> > I'd like to propose a compromise, if we can't get EXIF orientation reinstated in 5.4.2. Which is addition of a flag to use orientation or not. This way we can have both ways. But this would be an API change and likely be dis-allowed from the start.  I want to make sure that all Qt users know of the upcoming breakage in Qt 5.4.2 and are available to release their concerns, so that the Qt team knows the extent that people will be pissed off from the change. If it's just me, then fine, I will accept that, but anyone doing camera capture on 5.4.1 should be concerned about this change.
> > 
> > If in fact the Qt does decide to break the "proper" behavior ,(since it's going to be in 5.5 after all!) then Qt users need to be given a turn-key fix to support 5.4.2 in the interim. It isn't proper that we have to scramble for fixes for a x.x.Y release when we know it is temporary breakage.
> > 
> > Thank you and please chime in.
> > 
> > _______________________________________________
> > Interest mailing list
> > Interest at qt-project.org
> > http://lists.qt-project.org/mailman/listinfo/interest
> 
>



More information about the Interest mailing list