Tk Revitalization: Methodology - *** THIS FORUM IS ARCHIVED ***

Key Pages

As the theory goes, you can't improve what you don't measure. And we haven't been measuring much in a long time. ;-)

We're suggesting that the following process is a useful means towards making measurable improvements in Tk.

Identify a fairly small number of Key Examples, Tk apps that will be used to focus improvements
Post screenshots etc. of these apps on key platforms, and their code.
Critique these example, generating a list of specific problems that they have (see below)
Apply improvements, whether this be core changes, sourcing in new widgets, new option settings, etc.
Track how the changes affect the number/severity of problems, before/after screenshots, etc..

In other words, issues with look and feel are bugs, and should be managed like any other defect.

Why Key Examples?

Basically, it means everyone has a common view on the process and its results (i.e. improvements between iterations). It invites easy comparison. With a nice easily-managed set of key examples, they can be packaged up and downloaded so that everyone can easily get them running on their own systems. This will be a huge help when it comes to generating screenshots on different platforms, doing critiques, etc. Note these may well (and probably should) mimic some good examples of applications we want to be able to emulate in Tk.

Two Types of Problems

The first type of problem is where there is a look and feel problem in the application. The primary goal is to get rid of these. Obviously, what we're really after is look and feel problems caused by issues in Tk itself. To a lesser degree, we're also looking for things where Tk made it hard to design the right solution, and easier to design a crappy one. So these problems have to do with the usability of the resulting application.

The second type of problem is when things look right, but that the code or effort needed to get that look requires an amount of effort inconsistent with the supposed ease of programming with Tk. So for example, needing to tweak many options or spacing to get a standard widget, or needing a huge pile of code to generate tabs are both bad. These problems then are not with the usability of the resulting application, but with the usability of Tk for the programmer.