On 28 June 2015 at 21:17, Martin Gräßlin <mgraess...@kde.org> wrote: > This worries me a little bit. Not the fact of how much time is available, but > the fact that you do a rather short term thinking. Let's put it simple: there > won't be any bug reports prior to October when the first large distros start > to ship it. And then it's too late to go back. We need to be sure that the new > solution is not giving our users a hard time. > > Thus first in extragear to roll it out to some users for testing. Once we see > that it's a good solution we can roll it out to all users by making it > replacing KSnapshot. >
My point was that the moment we get any bug reports (even in October), I'm free enough to roll out bugfixes in a very short time. You may think I'm thinking short-term, but I've budgeted a lot of active development time for this at least to the middle of the next year. I have feature-addition plans, and of course there's a Wayland backend that has to be made and tested. While the KSnapshot code is taken to be "complete", I'm going to say that KSG's development state is "highly active", and we're practically at feature-parity with KSnapshot. KSG has a future. KSnapshot's current codebase, at least in the Wayland era, does not. Also, I've been using KSG instead of KSnapshot ever since I started building it, and I take a *lot* of screenshots. I know I come with confirmation bias, but I'm more than confident that showstopper bugs are not there. I can guarantee that you will be able to take screenshots, that you will be able to send them to other applications, that you will be able to save them (even to KIO remote locations - I've tested with mtp), that copy-to-clipboard works, etc etc. I've even tested running it on Weston with a do-nothing backend, and it starts and I can use the program (without taking an image) just fine, so I know that the Wayland/X11 backend decoupling is working just like it should. > >> >> This said, moving to extragear for the purposes of extra real-world >> testing is a good idea, except that unless we can give users a huge >> incentive to eschew using KSnapshot (which is in kdegraphics and hence >> released with the KDE apps, and receives preferential treatment from >> distro packagers over extragear apps), and install KScreenGenie >> instead, very few users are actually going to use KScreenGenie. > > This worries me. I think we should be happy with few users at the start for a > new product. > We'd want to start with a small userbase, if this was huge program with lots of new stuff. However, it's not. Most of the core functionality (KSCore.cpp) is just copied from KSnapshot and refactored. The native XCB-based X11 backend (which I take credit for writing) has been tested to death, and it works just fine. Things already work. The screenshot capture program is a core part of the desktop and also a thankless utility. No one pays much attention to it, but if it doesn't work people get very unhappy. I've taken extra care to make sure there are no "I can't take screenshots and save it" bugs. There are none. Whatever bug reports are expected are feature requests. Why I want a big user base is that I expect bugs that happen only on esoteric systems, and bugs that happen only on one system per every thousand installed. This kind of bugs I'm anxious to find and stomp out. > >> There's also the fact that once KScreenGenie makes it out in public >> using the KSG name, it'll be difficult to change the name to KSnapshot >> again, because we end up doubly confusing users. > > names are well just names. If we rename it later on to ksnapshot it's no > problem. For the users it will just look like a new ui. I doubt that our users > know it's called "KSnapshot" at all. They use the print key on their keyboard. > > Cheers > Martin Thanks, Boudhayan
