Skip to content

Latest commit

 

History

History
97 lines (75 loc) · 4.74 KB

RATIONALE.rst

File metadata and controls

97 lines (75 loc) · 4.74 KB
Raw

Rationale

A most frequent question I've been asked is "why not NeoVim?". Before starting vai, and more or less a month before the announcement of neovim, I put my nose in the vim source code to see if I could implement the features I wanted directly into vim. Here were my findings

  • It's completely in C, and not good C because of portability considerations and an old programming style. Adding any of the features I want to that is going to be complete hell.
  • vim as it is is not multithreaded. Adding multithreading to vim is going to be an ugly task which I don't want to perform. On the reason why I want multithreading is because I want the editor to perform operations in the background, such as checking for syntax, checking the git for changes, saving backups, running tests and marking them as green/red as I type.
  • vim does not use ncurses, and for many good reasons (ncurses is really bad), but if I have to render something I don't want to rely on its internal IO layer. I want to get up and running fast with a well known well documented interface.
  • vim is bloated. Some features are not really needed today (e.g. rot13). Others are overdesigned (e.g. folding). Others rely on external tools (e.g. ctags) that are a bit ancient in design and choke on unusual situations.
  • I want features such as intellisense like autocompletion, semantic highlighting, multicursors and so on. These are complex features that you don't want to implement in C.
  • Developing plugins in python would be much cooler than with a proprietary language that has no practical use outside of vim.
  • It takes ages to get a productive vim configuration. vim is missing a proper configuration panel and access to an online plugin repo as in sublimetext.
  • I wanted to have fun reimplementing a Qt-like library to learn about Qt itself.

I decided it was a lost cause, and I started writing my own editor. I laid out the foundation which is currently vaitk, and implemented basic stuff.

Shortly afterwards, Neovim was announced as a refactoring of vim. It addresses most of my points above. The problem is that, being a refactoring, it will still be in C and will grow again to become ugly and nasty.

I am not saying they are not amazing programmers and they are not doing an amazing job. I am saying that it's time to move on. It makes no sense to write something that complex and without a real need for system programming in C in 2014. Leave C to low level programming, and use high level languages for everything else. Maybe they will eventually reach the point of coming up with a lean C core where everything is moved to plugins, but for now they haven't released anything that can run. I do. They have 96 contributors on the github. I am alone.

The Neovim guys want to keep compatibility to vim, or better vim plugins. I am also trying to keep compatibility, but only limited to key bindings and general behavior. I am not planning to preserve plugin compatibility at all. In a sense we have different goals. The question is if their goals actually matter.

Another question I am asked is "Why not use eclipse/anything else if you want an IDE?". Eclipse is great software but it's an elephant that's just not for me. I generally find GUI based development slower, considering my terminal and keyboard oriented workflow. I would have to throw away and relearn years of muscle memory training, and it's an investment I am not willing to do. While vim compatibility modes do exist, I find them limited and I still don't get the flexibility I get from pure vim. I don't even use gvim, just to contextualize.

Development methodology

I develop in short feature-driven cycles of 2 or 3 weeks, trying to keep individual entries in the github feature within a proper scope so that I implement around 30 features per cycle (according to my current last three cycles).

I have other projects to attend in my non-working time, in particular

I jump from one to the others in a round robin pattern, occasionally giving more attention to the one that needs it the most depending on the circumstances.

If you want to help with development, there are plenty of small little things to do, even not requiring a lot of training or time, and every bit matters, but the fact that you are trying it out it's already of great help.

About vaitk

Vai foundation rests on VaiTk, a terminal UI toolkit heavily inspired by Qt. At the moment, VaiTk is part of Vai, but it will eventually become a product in its own merit.