A new folding editor

After working with Origami for many years, both as user and developer, I have decided to implement a new folding editor: fe. There are a few basic design decisions which make it different from Origami:

Origami is partially written in its extension language. While being more of a hack in the beginning, it has changed to a full programming language. As such, it takes time to be learnt. Code written in it is not well usable by most people who don't know it. The conclusion for fe is not to support any extension language, neither a homebrew one or a standardised language, like scheme. Instead, fe is implemented as a library of basic editor primitives. Its is easy to write your own editor by using that library. fe can be modified easy without having to learn a new programming language. The editor also stays small and elegant this way, while Origami has to offer zillions hooks for extension functions.

Origami implements folds as double-linked n-trees, with nodes being single lines or folds. This allows quick manipulation of folds or lines. But region oriented functions become hard to implement and are mostly not too quick any more. fe uses a gap buffer to store text, folds are represented as meta-characters in the flat buffer. This means basic fold primitives are slower than in Origami, but more complex and region oriented functions are faster. During development of the first prototype of fe, I found many functions in Origami not to be as canonical as they should be after I implemented them in fe. From my past experience, the performance of typical personal computers to have increased by a factor of 10 every 5 years in the last decade, so the circumstances for editor design have clearly changed over time.

Folding has its merits, because it adds a structure to flat files. But, it also means that by bad structure design and badly commented folds the file will get harder to understand, not easier. This can happen up to the point where you want all folds to be gone to see what the file contains. If you need to keep many folds open during development to see their contents, you are probably prepraring that situation at least for others. Although in general I don't like programs forcing me to do something, fe makes an exception here for two reasons: If the structure is obvious, you want the editor to close folds for you automatically. If it is not, you want to recognise that before it is too late. So, fe closes all folds, when you leave them. If you want to transport code from one fold to another, just split the current display and edit the file at two places.

Currently fe is a tiny editor. It implements folding, regions (including filtering and an Emacs-like kill ring), keyboard macros and splitting the screen in displays. There is only an Emacs-like user interface.

Michael Haardt