Irreal

Sacha Chua has posted a followup to her interview with Magnar Sveen. It’s basically a set of topic headings that index into the video of the interview. That’s very nice because if you want to revisit a particular part of the interview, you can go right to it rather than having to run through the entire video.

She’s also added audio only links in MP3 and OGG for those who want to listen during a commute. That’s very nice too.

Chua is famously “semi-retired” but the interview and followup represent a tremendous amount of work. We should all be grateful for her taking the time to do this and maybe leave a thank you on the post at her site.

Matthew Keeler has posted a nice introduction to writing YASnippets. Keeler starts with a minimal snippet that simply includes the GPL 3 boiler plate in the file and then moves on to a for loop snippet for C/C++ that uses tab stops.

Finally he presents a couple of complex examples. The first uses mirrored fields to build C++ QT classes. It also demonstrates embedding Elisp code in the snippet. A second example ties everything together to build a snippet for defining a PHP class.

Writing snippets is pretty easy once you’ve seen a couple of them explained. Keeler’s post is a nice way to get started.

Here’s my solution to the last EmacsGolf challenge. I used the apparently little-known how-many command. I normally use how-many in Elisp to count the occurrences of some regex. Here’s an example from my old blog that leverages some code from the always useful EMACS-FU. It turns out that how-many is a command and thus callable interactively. That makes the solution straightforward.

That does the job in 8 keystrokes.

The excellent flx-ido mapped hm to how-many on the first try despite the fact that I hadn’t used how-many in several months (and never before interactively). I map 【Meta+/】 to hippie-expand, which takes care of expanding the abbreviation.

After watching Sacha Chua’s interview with Magnar Sveen, I wrote that Sveen had demonstrated some Emacs features that I wanted to try out. One of those was flx-ido. The flx package was written by Le Wang to bring Sublime Text’s fuzzy matching to Emacs. Wang describes it as fuzzy matching with good sorting.

The ido package already supports fuzzy matching, of course, so the advantage of flx-ido is the improved sorting of prospective targets. You can see flx-ido in action with comparison searches using ido flex in this screencast from Wang. It nicely illustrates the advantages of flx-ido in a way that writing about it can’t.

I’ve been using the package for about 3 weeks and, after some acclimation, really like it. Initially, I used it like I did ido flex but that gave me suboptimal results. Once I started typing the first character of each word—some-fancy-command → sfc—I got excellent results. If you decide to try it, give yourself a bit of time to get used to it and train yourself to type appropriate abbreviations.

The other thing that I find a huge advantage is ido-vertical-mode. It lists your ido choices vertically instead of the default horizontal method. That makes it much easier to see the choices and pick the correct one. This was one of those things that I was pretty sure I wasn’t going to like but I got used to it right away and wouldn’t want to go back now. Even if you decide not to try flx-ido, do yourself a favor and give ido-vertical-mode at try. I think you’ll like it. You can see it in action in the Sveen interview or in an animated gif at its project page.

Here’s an easy EmacsGolf challenge that came up today. I have an org file in which some of the headings have a check mark (✓) at the beginning of the heading (but after the stars). The file is quite long and I want to count how many check marks there are in the file.

For clarity, the file looks something like this

 * Level 1.1
 ** ✓ Level 2.1
   blah blah blah
 ** Level 2.2
   blah blah blah
 * Level 1.2
 ** Level 2.1
   blah blah blah
 * Level 1.3
 ** ✓ Level 2.1
   blah blah blah
 ** Level 2.2
   blah blah blah
 ** ✓ level 2.3
   blah blah blah
...

Here are the rules:

1. There are no check marks in the body of the text, only at the beginning of some headers.
2. You may assume that there’s an abbreviation, cb, for ✓.

The challenge is to find the total number of check marks in as few a number of keystrokes as possible.