As I started reading the article, I began smiling thinking this guy is a Qt fananatic. Then wham he downplays Qt and picks GTK? Get real.
Here's my 2 cents.
The design of the GUI only benefits the designers. (Please no flames on that!) As a programmer useability (of the GUI) is probbably the highest thing on my plate and that goes way WAY beyond GUI design. For instance, in a purely get the work done framework, I recently switched a project from C++ Builder to VB. Do I like VB? Absolutely not. But it has one advantage I have yet to see in ANY other development environment. I need to build a lot of screens really fast, really robust, with lots of code under them to ensure they did the right thing at the right time. My first attempt at creating a wizard for all the maintenance stuff resulted in 3 lines of code to create a form and place a control. 4 lines of code let me reference a control that wasn't part of the design environment by default. The wizard I wanted, just required prototyping the 6 screens I needed for doing maintenance on a single table. I often find that useless features get tucked into programs, for instance, controls that access data and allow updating in Windoze. I haven't worked on a single toolkit that should have integrated data access into controls for anything other than a read-only envrionment. -Give me a break- I have NEVER gotten intrinsic data access to work for any but the most trivial of examples and those are often in deep question of whether they work as they should. Bottom line, fewer "features", higher flexibility, higher extensibility, and a great environment (that it is trivial to extend, and that doesn't mean the extensions have to be trivial)
The next issue which has to be fairly high on the list has to be speed of the resulting application. VB is NOT a speed demon, but I can't type faster than it takes characters, that's fast enough.
On a Linux platform I'm still looking for something worth developing in. FLTK is a great niche tool (Because it's so darn small!) Beyond that Qt is my next love/hate relation ship. The library (design) IMHO is great. Not nessasarily because it's an elagant design, but because it's consistant and easy to understand. My hate relationship is the license. It would be nice to see a graduated license.
GTK certainly doesn't fit the bill for me. How exactly you correct error free code in a consistant manner that's easily understood and enforced by the compiler is still eluding me. (But hey, the docs aren't quite done yet are they:))
Bottom line, the GUI design really only affects the GUI designers. I could care less if the developers sweat bullets to create a fantastic GUI to develop with (unless of course I'm writing the GUI!). Programmer and end-user useability should be more important than the beauty of the design. Besides the most elegantly designed GUI used by 3 people isn't likely to create a fundamental shift in Global programmers lives. VB is ugly but easy to use and it gets used by millions of developers (some use it just to prototype what they will develop in another language). Make the environment and the GUI trivial to use and trivial to learn and it will get used, a lot. Don't add features because you can but because they are useable. Work more on diversity of implementation (Controls, etc) than intrinsic control features. A great GUI will make writing a Game GUI a snap; a system utility a breeze; and a productivity app a joy.
Slashdot Mirror
As I started reading the article, I began smiling thinking this guy is a Qt fananatic. Then wham he downplays Qt and picks GTK? Get real.
:))
Here's my 2 cents.
The design of the GUI only benefits the designers. (Please no flames on that!) As a programmer useability (of the GUI) is probbably the highest thing on my plate and that goes way WAY beyond GUI design. For instance, in a purely get the work done framework, I recently switched a project from C++ Builder to VB. Do I like VB? Absolutely not. But it has one advantage I have yet to see in ANY other development environment. I need to build a lot of screens really fast, really robust, with lots of code under them to ensure they did the right thing at the right time. My first attempt at creating a wizard for all the maintenance stuff resulted in 3 lines of code to create a form and place a control. 4 lines of code let me reference a control that wasn't part of the design environment by default. The wizard I wanted, just required prototyping the 6 screens I needed for doing maintenance on a single table. I often find that useless features get tucked into programs, for instance, controls that access data and allow updating in Windoze. I haven't worked on a single toolkit that should have integrated data access into controls for anything other than a read-only envrionment. -Give me a break- I have NEVER gotten intrinsic data access to work for any but the most trivial of examples and those are often in deep question of whether they work as they should. Bottom line, fewer "features", higher flexibility, higher extensibility, and a great environment (that it is trivial to extend, and that doesn't mean the extensions have to be trivial)
The next issue which has to be fairly high on the list has to be speed of the resulting application. VB is NOT a speed demon, but I can't type faster than it takes characters, that's fast enough.
On a Linux platform I'm still looking for something worth developing in. FLTK is a great niche tool (Because it's so darn small!) Beyond that Qt is my next love/hate relation ship. The library (design) IMHO is great. Not nessasarily because it's an elagant design, but because it's consistant and easy to understand. My hate relationship is the license. It would be nice to see a graduated license.
GTK certainly doesn't fit the bill for me. How exactly you correct error free code in a consistant manner that's easily understood and enforced by the compiler is still eluding me. (But hey, the docs aren't quite done yet are they
Bottom line, the GUI design really only affects the GUI designers. I could care less if the developers sweat bullets to create a fantastic GUI to develop with (unless of course I'm writing the GUI!). Programmer and end-user useability should be more important than the beauty of the design. Besides the most elegantly designed GUI used by 3 people isn't likely to create a fundamental shift in Global programmers lives. VB is ugly but easy to use and it gets used by millions of developers (some use it just to prototype what they will develop in another language). Make the environment and the GUI trivial to use and trivial to learn and it will get used, a lot. Don't add features because you can but because they are useable. Work more on diversity of implementation (Controls, etc) than intrinsic control features. A great GUI will make writing a Game GUI a snap; a system utility a breeze; and a productivity app a joy.