Is SaaS Killing Native Linux App Development? 330
jfruhlinger writes "In a world where 'app' is the new buzzword, the development of native Linux apps is lagging. Some of this can be attributed to the usual community infighting (the latest version of which is argument about Ubuntu's Unity interface), but there may be something deeper at play: Linux advocates have for so long advocated browser-accessed software as a service as a way to break out of Microsoft's proprietary desktop. Now that this world has arrived, there's less incentive to work on native Linux apps. But of course, entrusting your functionality and data to a cloud provider like Google has its own set of concerns for free software fans."
Re:Native GUI app development is a pain (Score:3, Interesting)
Nonsense! Qt Creator is a perfect counterexample!
*Easy to use: Check
*Widely supported: Check
*Actively maintained: Check
*Designed for use with a somewhat mainstream language: Ch... well... ahn... wait, but what is a "somewhat mainstream language" anyway?
Or, maybe Linux is dying... (Score:2, Interesting)
Linux is dying (or perhaps dead). At least it is on the desktop. It was stillborn and never really had a chance. Everyone (myself included) spent so much time worrying about Microsoft that we ignored what Apple was doing - and then *wham*, OS X comes out. A Unix-backed desktop OS with a gorgeous UI that non-techies can actually use.
Without a strong desktop presence, there is very little need for native apps. We sysadmins prefer our command line tools - and nobody can argue that there aren't a lot of those ... but general app development on Linux dropped off years ago and I don't see it improving anytime soon.
And you know, I really don't have a problem with that. I started with Linux in the early 90s, with one of the first releases of Slackware. Back then monitors were fixed-frequency and you had to calculate your video card's dot-clock & other timings in order to not blow up your monitor... I became a full-fledged sysadmin in 1995 and worked for a number of big Linux companies. I drank the kool-aid... all my machines at home ran Linux and it was good. In 2000 I switched gears and became a Linux developer - working on both embedded and desktop projects. Had some great experiences back then.
But somewhere around 2002 I started to feel betrayed. Here I was, nearly 10 years later and Linux still wasn't on the desktop - at least not in any kind of meaningful way. Sure I kept hearing how 1997...1998...1999...2000...2001 were all going to be the "Year of Linux on the Desktop" - but it never happened. The various Linux forums were the same old thing - people complaining about Microsoft (and now Apple) - all the while lifting as many UI ideas as they could from each OS, expecting that someone a floaty OS X dock-like thing would attract hundreds of thousands of new desktop users to the platform.
Here we are in 2011 and I'm seeing the same old shit. I'm just about ready to give up Slashdot because 90% of the Linux-related news stories just remind me that we haven't made any progress (and yes, as a developer I tried to help out in that area). Ubuntu's now saying that they're going to make a tablet UI. Yay ... only they haven't cracked the desktop (they made a lot of progress and undid it all with Unity).
*sigh*
Run the server locally (Score:4, Interesting)
HTML5 is just another GUI front-end library. In no way does it require you to write cloud based apps. If you want a native Linux application write the GUI in HTML5 and run the server on the same machine as your GUI. Hmmm.... something kind of like the Xserver model, but brought 30 years into the future?
When people whine about the ending of location transparency with the Xserver, what is going away is the Xserver as the primary GUI library, not location transparency in general. The Xserver needs to die, it is pass its prime and we need to move onto newer GUI technologies.
So stop writing native Linux applications and instead start writing HTML5 applications that ship with a built-in server. The cool thing about apps in this model is that the GUI works on Linux, Mac and Windows plus you can run the server locally or in the cloud - your choice. If you want to help out convert some native Linux apps into the HTML5 model.
Wayland is a key transition technology. It allows apps like Chrome/Firefox to be written directly to EGL. Plus you can run a user space Xserver as a legacy tool.
hmm (Score:4, Interesting)
Re:Native GUI app development is a pain (Score:5, Interesting)
Qt is an incredible toolkit. The only problem is that it looks "different" in Gnome, but about a year ago this problem seems to be solved.
Qt Creator is one of the highest-quality IDEs - very easy to use, powerful and not getting in the way.
Plus as a bonus Qt apps can be easily ported to Windows and MacOS, especially if they aren't using anything outside the Qt toolkit - many apps will simply compile and run with zero changes. Qt includes stuff like XML parsing, sockets, OpenGL etc. so you can probably will never need anything except Qt.
The only thing some people don't like about Qt is the need for a preprocessor and duplication of C++ stdlib stuff (like containers, I/O operatuins), which supposedly fragments C++ development. But I actually like this - no need to use use multiple library dependencies, everything is included in Qt SDK, along with great documentation and works out of the box.
There are concerns about Qt's future, since it no longer fits in Nokia's strategy. But it's quality definetly exceeds most toolkits like wxwidgets, gtk, mono and many others. Probably because the "boring" stuff like documenting and testing was done by full-time employees.
Oh, and Windows native C++ development is horrible. Just look at how you're supposed to display the standard "open files" dialog and get the selected file names.
Re:Native GUI app development is a pain (Score:4, Interesting)
Oh, and Windows native C++ development is horrible.
I have a stupid question... What is actually C++ about Windows Native Libraries. All I ever saw was just completely twisted C - twisted as in just weird datatypes, weird includes, and a Macro hell. I admit, you can declare variables anywhere (just not on top, as in C), but otherwise I cant understand how Microsoft can call it C++.
Real C++ is so different from Microsoft C++.
And QT is very nice. It combines the best of C, C++ and Java. QT Creator is the only IDE I ever appreciated. I really like the .pro file that contains everything about the project, and is still a very small simple text file; nothing like a configure-script or even worse a Microsoft Solution file. qmake uses the pro-file to make a Makefile.