Comparing IUP with Other Interface Toolkits
Why to still maintain IUP if today we have so many other popular toolkits?
This is a question we always ask to ourselves before going on for another year.
To answer that question we must first define the characteristics of the "ideal" toolkit, list the available toolkits and compare them with the "ideal" and with IUP.
We would like a toolkit that has:
- Portability. That provides an abstraction for User Interface in Windows, UNIX and Macintosh.
- Free License and Open Source. This means that we can also produce commercial applications. The pure GPL license can not be used but the LGPL can but must contain an exception stating that derived works in binary form may be distributed on the user's own terms. This is a solution that satisfies those who wish to produce GPL'ed software and also those producing proprietary software. Many libraries are distributed with this license combination.
- Small and Simple API. This is rare. Many libraries assume that an Interface toolkit is also a synonymous of a system abstraction and accumulate thousands of extra functions that are not related to User Interface. At Tecgraf we like many small libraries instead of one big library. Almost all available toolkits today are in C++ only, so C applications are excluded, also this means a hundred classes to include and understand each member function. The use of attributes makes a lot of things more elegant and simpler to understand.
- Native Look & Feel. Many toolkits draw their own controls. This gives an uniformity among systems, but also a disparity among the available applications in the same system. Native controls are also faster because they are drawn by the system. But the problem is what's "native" in UNIX? Some commercial applications in UNIX start using Motif as the "native" option. It is the official standard but because of license restrictions, before the OpenMotif event, the system became old and some good alternatives were developed, including GTK and Qt.
Toolkits
With these characteristics in mind we select some of the available toolkits:
Name |
License | Last Update | Version | Language | Platforms | Controls | Team | Comments |
LGPL* | 1997-2006/08 | 1.7.4 |
C++ |
Win, X |
own |
3 | great look, license restrictions | |
LGPL* | 1998-2006/01 | 1.1.7 |
C++ |
Win, X, Mac |
own |
4 | was from Digital Domain. Easy to learn. |
|
LGPL* | 1997-2006/07 | 2.10.1 |
C |
Win, X, Mac |
own |
9 | target for X-Windows, basis of GNOME, Windows is apart, Mac using X |
|
GPL | 1994-2006/06 | 4.1.4 |
C++ |
Win, X, Mac | own | (many) | Is free for Non Commercial, basis of KDE, Emulates the native look and feel | |
LGPL* | 1992-2006/08 | 2.7.0 |
C++ |
Win, X, Mac | native | 6 | X can use Motif or GTK | |
MIT* | 1994-2006/09 | 2.5.1 | C | Win, X, Mac | native | 2 |
X can use Motif or GTK, Mac using X |
Table Last Update: August 2006
More toolkits can be found here:
The GUI Toolkit, Framework Page.
An interesting article can be found here:
GUI Toolkits for The X
Window System.
Conclusions
From the selected toolkits using the defined approach we can eliminate some toolkits:
The gray ones are not updated anymore or the development is very slow or needs a better organization.
FOX has a great look but the license can be restrictive in some cases.
FLTK promises a new version with a better look, but until then it does not have a pretty good look. The FLTK documentation also does not help.
GTK+ can be used as a replacement for Motif, but not as a fully "portable" toolkit since is was target for X-Windows. Nowadays GTK+ 2 is a great free C toolkit. But some predefined dialogs could be the native ones, like the File Selection, specially in Windows. The Windows port works ok but it is not Windows behavior neither look and feel.
Qt has several license limitations, although is a very stable and powerful toolkit. Qt can be also used as a replacement for Motif.
The "best" free solution that we choose would be wxWidgets because of the native controls and its portability. But since version 2, GTK+ is a very strong option because it is in C and had its visual improved.
It is very hard to compare IUP with wxWidgets and Qt since they are much more than an Interface Toolkit. They are complete development platforms that include several secondary libraries not related to User Interface. In IUP we focus only in Graphical User Interface.
Developing IUP
IUP has a C API, only has functions for Graphical User Interface, and uses "Native Controls" in Windows, Motif and GTK+. These are the major differences between IUP and other toolkits. Because of that IUP is small, fast and very powerfull.
We have a small but very active team and we have many Tecgraf and foreign applications that today use IUP, collaborating for its evolution. Our objective is to surpass the Tecgraf needs, keeping backward compatibility and improving the internal code.
IUP does not have a wide localization feature, it only includes support for messages in English and Portuguese. And it does not have support for Unicode characters.
Why Not Mac? The first Mac driver was developed for MacOS 9 and had several memory limitations so it was abandoned. With Mac OS X we have the opportunity to do something better. Today IUP runs on Mac OS X using X11 and Motif or GTK. We plan for the future to build a native Carbon driver.
Why Motif? Motif is still very important for non Linux systems, some Tecgraf applications run on AIX, SGI and Sun systems, that only have Motif installed and we can not force the installation of other toolkits like GTK.
.. "Make it Reusable, Make it Simple, Make it Small" ...