Preface

Why, oh why, would anyone be interested in Leo? After all, Emacs and Vim are superb text editors, and Visual Studio and Eclipse are great IDE’s. How can Leo possibly compete with such strong competition? What does Leo offer that these other tools don’t?

Leo does have something unique to offer–something missing from Emacs,Vim, Visual Studio and Eclipse. Leo’s users often speak of an Aha! moment when they understand what this ‘something’ is. The Aha! arises from considering programs, design and data in a new light. You might call this ‘The Leo Way’. In essence, Leo shows that computer programs, designs and data are not, after all, mostly about text. Yes, people usually express programs, designs and data as text. Yes, people manipulate programs, designs and data using text, but text is not the whole story.

In The Leo Way, text is simply a manifestation (a shadow) of something more fundamental. That ‘something else’ might be called ‘organization’ or ‘structure’ or ‘views’ or even ‘architecture’. However, let us use the term node to represent the fundamental unit of structure in Leo. We’ll see why in a moment.

In architectural terms, nodes are bricks that make up a building. In computer programming terms, nodes make up methods, classes, files and entire applications. So the term ‘node’ does not have a fixed meaning–it is simply a unit of organization. Any node can be built from other nodes, and any node can be used by any other node. Leo represent nodes directly, as nodes (headlines) in an outline. An outline node contains a headline and body text. The outline pane shows all headlines; the body pane shows the body text of the presently selected node.

Outline structure is real data

To repeat: the fundamental unit in Leo is not text. True, headlines and body consist of text, but a node is just not text, it is a true (Python) object. This means several specific things:

  1. Because nodes are true objects, Leo commands understand what a node is, and where a node fits into the entire outline. I’ll say more about outline organization soon, but let me give an example. Every node has exactly one parent node, (except for top-level nodes that have no parents) and every node has zero or more children and zero or more siblings. A node’s parent, children and siblings are real properties of the node, completely independent of the node’s headline or body text. Furthermore, any of Leo’s commands (or user-written scripts or plugins, the big sisters of user scripts.) can easily access the all aspects of an outline without having to parse any text whatsoever. Commands, scripts and plugins can easily do the following: get the root of the outline, the presently selected node in the outline, the parent, siblings, children or descendants of any node in the outline, etc., etc. Commands, scripts and plugins can easily insert, delete or move nodes, and can alter the headline or body text in any node. All this without parsing text.
  2. Having nodes be true objects means that commands scripts and plugins can treat the headline text as something truly different from body text. The natural interpretation of headline text is as a description of the body text. This is important! Headlines often control Leo’s commands. For example, headlines that start with @file, @asis, @auto, etc. serve to guide and control Leo’s read and write commands. Headlines that start with @test, @suite and @mark-for-unit-tests guide Leo’s unit testing commands. Moreover, it is easy to create new conventions for headlines that control user-written scripts or plugins. For example, plugins define specific meanings for headlines that start with @url, @rst, @bookmark, @slideshow, etc., etc. So the separation of headline and body text, as true components of a node object, is a very big deal.
  3. One application of these ideas deserves special mention. Leo’s scripting plugin provides support for @button nodes. The headline is @button <command-name>. The body text contains a script. When Leo opens a Leo outline, each @button node creates a command and an icon. Clicking the icon (or executing the command) applies the script in the @button node to the presently selected outline. That is, the script is executed in a context in which it is easy to get access to all aspects of the outline in which the script is embedded. This is a major advance in scripting. It allows you to ‘bring scripts to data’, i.e., any part of an outline. In particular, it is very easy to create editing scripts that automate what would otherwise be boring and repetitive editing tasks.

Leo outlines aren’t your average outline

Earlier I said that any node can be built from other nodes, and any node can be used by any other node. It takes a very special kind of outline for this to be possible. In a typical outline, such as Emacs outline mode, for example, nodes appear exactly once in the outline. This makes it impossible to ‘reuse’ nodes in multiple places. Leo removes that limitation: any outline node can be cloned, and clones can appear in as many places in an outline as you like. Although clones may look distinct on the screen, at the data level each clone is exactly the same node.

Earlier I said that you can think of nodes as representing ‘organization’ or ‘structure’ or ‘views’ or even ‘architecture’. Clones are the crucial feature that allows this point of view. For example, we can build up multiple ‘views’ of data in an outline using clones as follows:

  • Create a ‘view node’ that will represent a user-specified view.
  • Clone all nodes that are to be part of the view, and move them so that each clone is a child of the view node.

That’s about all there is to it. The view node, and its children is a new view of the outline. This notion of ‘view’ is so important that Leo supports it directly. Leo’s chapters are simply views created as I have just described. When you select one chapter, you only see the nodes of that chapter in Leo’s outline pane.

Conclusions & encouragements

So Leo offers a new way to understand, organize and manipulate any kind of complex data, including computer programs, designs of computer programs, web sites, personal data, whatever. The Aha that I invite you to experience is this: Outlines are more than mere eye candy. Having organization be real data creates an entirely new dimension, literally and figuratively, in computer programming, computer design and data organization, including web-site design, database design, etc. Leo’s commands use headline and body text in many creative ways. So can you and your scripts. It’s easy, it’s fun, and it’s revolutionary.

That’s about it, except for some words of caution and advice:

  1. Leo has been under active development for over 10 years. The new world created by nodes is rich and varied. You won’t learn it all in a day or so. Please be patient. Start by learning Leo’s basic features as explained in the tutorial. You can learn more advanced features later.
  2. Those of you who are comfortable with Emacs should feel pretty much at home with Leo. Leo has shamelessly stolen the best features of Emacs, including the minibuffer and many Emacs-like commands.
  3. For those of you who are not comfortable with Emacs, please understand that you do not need to understand all of Leo’s commands in order to use Leo. Start by ignoring the minibuffer. Later, the minibuffer can become your friend, but you can get the Aha! without it.

Edward K. Ream July, 2007

Previous topic

前言Front Matter

Next topic

What People Are Saying About Leo