Yeah API between major version is going to change, because, yeah the world is evolving and better concept become available, if you thing all the API changes are for idevice only, your saying but I beg to differ. I really like where they are heading. QWidgets is hard to maintain and is one reason it's mostly left as is, because it require a lots to maintain the native GUI on each platform so new platforms require a lot of effort to be added (this is no more a 3 OS only world anymore, those days have been over, move on).
Qml (I have made 3D medical CAD application desktop, and it run flawlessly), I for one have ditch QWidgets and I don't plan to look back to the pain of connecting everything by hand, error prone and make way to many hard to maintain code. Qt bindings is coming to C++ and this will also be a huge welcome from me. The changes that was made is exactly to adapt to the world, and multiple platform out there. Qml also open the door to Qt for MCU, to Web assembly ... I fail to see why the iDevice is a concern, they are expending the allowed paltforms and this is good for a cross platform framework to do so. If your developer have hard time learning Qt3 to Qt5, I feel sorry for you, your team is probably weak (but this is my personnal opinion). They must have a hard time with C++11 to C++17 too. I have port application from Qt3 to Qt4 back in the days, from PPC to x86. From Qt4 to Qt5. The new code mostly felt cleaner in the end. Nobody is force to update from major version if that suite you. If you used an unsupport OS, you probably still used the lastest framework that were designed to support those, don'T expect Qt to support OS that aren't event supported by the manufacturer anymore. You cannot take such a told in the long run. But if you feel like the project has gone side way and you prefer the alternative fork, this is what open source is for. -->1) Kids who only learn the new stuff can't be hired to support the old because they cannot even begin to function. - We do have people from 22 to over 50, they code from embedded mcu C to Qt5 without problems, I suggest you hired from different schools. --> 2) If someone decides to pull the plug on their really old embedded OS in favor of a roll-your-own Yocto Linux build, they can't even bring their code forward. The entire cross-platform aspect of Qt has been abandoned. It's only cross-platform for "the cool kids." - Far from it, I found Qt more open to platform (MCU, embedded, web...). What is wrong with rolling out Yocto? for embedded system this is a good way. If the OS is deprecated and unsupported, continue to use the tools that were going along with it and stay frozen into time with it if you want. -->*No core API changes that break backward compatibility. * - They mostly have minor changes between minor version or not even at all. Those are between major Qt version and let's face it, it's normal to make thing evolve, just like C++ is evolving too. I manage to maintain few application from 5.6 to 5.15 in the past few years, the most problems I got was with Android/iOS and the Play/App Store requirements, the Qt was easy to and mostly smooth. My opnion, doesn't reflect any cie but my own. -----Original Message----- From: Roland Hughes <rol...@logikalsolutions.com> Sent: June 15, 2020 12:06 PM To: Jérôme Godbout <godbo...@amotus.ca>; interest@qt-project.org; Thiago Macieira <thiago.macie...@intel.com> Subject: Re: [Interest] [Development] Windows 7 support will be dropped in Qt 6 You completely miss the point. Qt has been prone to sweeping API changes mostly due to the fact they keep chasing the iDiot phone market. If that is the market they want, fine, but they had best fess up now so everyone else can go somewhere else. You can't take someone who learned via Qt 5 and have them work in Qt 3 because it is night and day different. You can change an awful lot without going down the full clinical trial path. IT ALL DEPENDS ON RISK. That device I linked went through some kind of enhancement approval path. I don't know what it replaced exactly, but I do know the previous device did not use Linux or Qt. I know because I worked on that project. When you get out into the world of industrial controls, they don't have the FDA level of certification. They do have 20-50 year lifespans though. These are things with a base price of half a million going well up past $5-$6 million. They have a minimum 30 year life span, most are closer to 50. Many of them are running DOS, WinCE, and the embedded version of Windows 7. It's bad enough that Qt keeps dropping everything the industrial and medical world needs in pursuit of the iDiot phone market, BUT ADDING INSULT TO INJURY THEY KEEP MAKING SWEEPING API CHANGES. This means two things: 1) Kids who only learn the new stuff can't be hired to support the old because they cannot even begin to function. 2) If someone decides to pull the plug on their really old embedded OS in favor of a roll-your-own Yocto Linux build, they can't even bring their code forward. The entire cross-platform aspect of Qt has been abandoned. It's only cross-platform for "the cool kids." What we really need is for KDE to announce what their new development library is. The desktop, medical, and industrial device world will move their money and resources to that library and Qt will be left with nothing but the phone market. The "new" choice must promise the following: *No core API changes that break backward compatibility. * They can add optional parameters to the end of a parameter list, but no class name changes, no method name changes, no dropping of methods. They can add new classes, but everything that was there in 2019/2020 has to still be there in 2050. The insult added to the injury of Qt dropping operating systems is they change sh*t willy-nilly without even the slightest thought given to the installed base. The people who aren't starting over every six months. On 6/15/20 10:10 AM, Jérôme Godbout wrote: > This is exactly my point, that device is STILL on Qt3 because you don't want > to go all over that certifications/testing just for changing a Qt versions. > You can enchance it, but you will need to proof the changes impacts have been > cover in testings and no risk have been added. Those system, nearly never get > update libs or anything at large because you have to retests and recertify > the whole thing. Just changing the model of an harddrive into a machine make > a need to recertify. So those device will never care if the bleeding edge Qt > doesn't support their platform. They are fixed with known version of > differents libs that are known to be workign together and tested. > > Those device are not using anything bleeding edge, so let alone runing any > Windows 7 or Windows 10 with Qt6 in the near time. If anything into those > domain start using Qt6 it will probably be a new design or a new version that > will need to be certified all over again anyway, so the upgrade path is not > there. > > -----Original Message----- > From: Roland Hughes <rol...@logikalsolutions.com> > Sent: June 15, 2020 10:11 AM > To: Jérôme Godbout <godbo...@amotus.ca>; interest@qt-project.org; > Thiago Macieira <thiago.macie...@intel.com> > Subject: Re: [Interest] [Development] Windows 7 support will be > dropped in Qt 6 > > I seriously beg to differ. > > In America "upgrades" and field patches have a completely different > certification path than "shiny new device." What has to be certified is based > on the extent of the changes and how well the FDA documentation is filled out. > > That OS/2 Qt3 medical device I get an email about ever 18 months or so > continues to get enhancements. Changing to even Qt for or Linux would push it > to "shiny new device" depending on the assessed risk level. > > A patient monitor like this one > > https://www.welchallyn.com/en/products/categories/patient-monitoring/v > ital-signs-devices/connex-spot-monitor.html > > Has a completely different risk level than say, an infusion pump. > > I have worked on products that just added new features to existing lines. In > America it happens all of the time. > > > On 6/15/20 8:33 AM, Jérôme Godbout wrote: >> I have work for medical devices for over 10 years and used Qt from 4.x to >> 5.8 (move out to IoT lately), designing system and software. Cie who do >> that, did it wrong, you have to ensure your software will run and you >> maintaint it, but in no way you will add any new features (you will need to >> certify again!). You keep a system images that can recreate the exact same >> output (OS, build tools, ...), you patch the bug that's all you should do. >> New features will be done into a new system that will need to be certify all >> over again. >> >> Upgrading to Qt6 for already certified devices is a no go, no matter what >> for them. This is totally irrelevent. You only upgrade tools and libs when >> you create a new system (version, design, etc) that will be certified again. >> >> >> -----Original Message----- >> From: Interest <interest-boun...@qt-project.org> On Behalf Of Roland >> Hughes >> Sent: June 13, 2020 11:08 AM >> To: interest@qt-project.org; Thiago Macieira >> <thiago.macie...@intel.com> >> Subject: Re: [Interest] [Development] Windows 7 support will be >> dropped in Qt 6 >> >> Medical devices are certified with their manufacturing process. >> Certification of something like a surgical robot can take 5+ years of >> clinical trials. That is _after_ you have done all of your internal >> development and cadaver trials. >> >> On 6/13/20 5:00 AM, interest-requ...@qt-project.org wrote: >>>> That's partially for their own peace of mind and stability, but >>>> along with that, many tool vendors take quite a while to certify >>>> their offerings, both hardware and software, which gives people >>>> another reason to stay behind. >>> More than two years? >> -- >> Roland Hughes, President >> Logikal Solutions >> (630)-205-1593 >> >> http://www.theminimumyouneedtoknow.com >> http://www.infiniteexposure.net >> http://www.johnsmith-book.com >> http://www.logikalblog.com >> http://www.interestingauthors.com/blog >> >> _______________________________________________ >> Interest mailing list >> Interest@qt-project.org >> https://lists.qt-project.org/listinfo/interest -- Roland Hughes, President Logikal Solutions (630)-205-1593 http://www.theminimumyouneedtoknow.com http://www.infiniteexposure.net http://www.johnsmith-book.com http://www.logikalblog.com http://www.interestingauthors.com/blog _______________________________________________ Interest mailing list Interest@qt-project.org https://lists.qt-project.org/listinfo/interest