Irreal

The other day I wrote about Álvaro Ramírez’s DWIM Shell Commands package. The TL;DR is that it’s a template framework for executing shell commands from within Emacs when the shell command has tricky syntax or hard to remember options.

Since writing the package, Ramírez has been revisiting some of his custom Elisp functions that call shell commands and refactoring them to use his new package. The first example is password protecting one or more PDFs. This example is noteworthy for how much shorter it is using the framework than it was originally. As Ramírez says, the fact that it’s so short and easy to write a custom function for invoking a shell utility means that he’ll do it more often and therefore streamline his workflow.

The second example involves converting a png file to a series of icns files. This example is distinguished by the fact that it’s more like a script than a single invocation of some utility.

You could, of course, do all this with a shell script or some custom Elisp but using the DWIM-shell-command package makes it easy because it does a lot of the work for you. If you find yourself invoking utilities via the shell command from Emacs, you should definitely take a look at Ramírez’s package.

I rarely use any of the diff utilities these days and when I do it’s mostly Ediff to resolve merge conflicts. That probably means I’m working too hard but I’ve always been that way: I just never think of using diff unless I’m really in trouble.

Most other folks are more reasonable and use some sort of diff regularly. A common use case is checking the differences between two versions of a file stored in a Git repository. Magit, of course, has that built in—several options, in fact—but Kaushal Modi really likes the syntax highlighted diff he gets with the command line tool delta.

Of course, Modi wants to stay in Emacs and use Magit. Fortunately, Dan Davison, the author of delta has a package, magit-delta, that allows it to work with Magit. Installation is pretty easy. Just install delta from your favorite package manager (it’s called git-delta in most package managers) and then install magit-delta by adding it to your init.el.

There is, apparently, an issue with displaying line numbers with git-delta (an issue has been raised) but it’s easy to work around that. Take a look at Modi’s post to see screen shots of both the normal delta and the magit-delta output. If you like what you see, just install delta and magit-delta and give them a try. It’s easy to turn it off if you decide you don’t like it.

UPDATE [2022-07-10 Sun 13:09]: Added link to Modi’s post.

Álvaro Ramírez is relentless in chasing down ways to eliminate friction from his workflow. One day it occurred to him that invoking shell commands was an area with plenty of opportunities for workflow optimization. A particular problem was that many of the commands he used frequently, such as ffmpeg and convert have a lot of arguments and flags that are hard to remember.

Ramírez’s solution was a write tiny bits of Elisp that would invoke the desired commands without having to remember all the details. To make this work, he first defined a general template framework that allowed specifying what the command should look like with parameters for the file path and parts of the file name.

Because it’s DWIM, the framework does the right thing depending on the context. If you’re in a Dired buffer, the command will operate on all the marked files. If you’re in a file buffer, the command will operate on the associated file. Similarly, how the output of the command is shown depends on the context in which it was invoked. Take a look at Ramírez’s post for the details.

You can check out the framework code, dwim-shell-command.el, to see how things work but the examples in the post are really all you need. Once the framework is in place, adding new DWIM commands is simple. The post has several examples.

If you’re the type of Emacs user who’s always looking for ways to sand down the rough spots in your workflow, take a look at Ramírez’s post. As Ramírez says, if your needs are different, it’s easy to adjust things to suit your requirements.

Torstein Krause Johansen runs Emacs in terminl mode on Linux. He likes to have several terminal sessions and switch between them. Previously, he used tmux to multiplex those terminal sessions onto a single screen but then tmux started crashing on him and he didn’t have the time or inclination to track down the problem.

Instead, he realized that he could do everything he wanted from within Emacs. That had the advantage of using the same key bindings that he was already using in Emacs and, less importantly, he didn’t have to worry about keeping the Emacs and tmux color schemes in sync. To accomplish that, he use vterm, multi-vterm, the built-in Emacs shell command, and a bit of custom Elisp, that he calls spawn-shell that allows him to create shells in named buffers the way he would in tmux.

Johansen has a video that explains how he does all this using just the packages and code above. It’s a really nice set up for those who like to run Emacs in terminal mode, of course, but even if you’re running Emacs in GUI mode, you can use the same strategy to provide multiple named shells all running from within Emacs. That’s pretty much what I do and it works well for me.

The video is short, only 5 and a quarter minutes so you can watch it pretty much any time you like. The slides are available but probably not too useful unless you want to copy the spawn-shell code. Definitely worth a few minutes of your time.