Irreal

The videos from the Emacs Conference 2019 are up at YouTube. As many of you know I’m a big fan of Perry Metzger’s 2014 talk on 31 years as an Emacs user so the first video I watched was what amounted to a follow up to that talk. In Emacs: The Editor for the Next Forty Years, Metzger looks at the future of Emacs, where he thinks it’s going, and what it needs to do to remain the premier editor that it is.

As Metzger says, it’s not 1992 anymore and many of the younger hackers aren’t adopting Emacs so we need to update it to meet modern needs. Of course, as Metzger also notes, Emacs has undergone continuous development and modernization throughout its lifetime. The question then—and the theme of the talk—is what changes do we need to make to Emacs? Much of what he has to say will probably ruffle some feathers or at least be controversial but whether you agree or not his arguments are well-formed and I found them cogent.

One of Metzger’s major points is that whatever we do, it must be done incrementally. One of the problems that every proposal for significant rewrites of Emacs has to face is the fact of the huge amount of existing code written for the Emacs ecosystem (Org-mode alone is 120K lines of code). That implies, for example, that implementing a new extension language means that it must interoperate seamlessly with Elisp and use the same run-time system.

Metzger’s summary at the end of the talk lists what he thinks the most important issues are:

• HTML rendering
• LSP support built into Emacs
• Modern email and PIM support
• A safe concurrency model
• A new extension language

The body of the talk builds the case for each of these changes and discusses how they might be achieved.

In a second video, Metzger takes questions on the talk. The questions are also worthwhile and expand on the talk. The two videos are 58 minutes and 23 minutes long so you’ll definitely need to set some time aside but if you care about Emacs this is an essential talk. The slides from the talk are also available.

Eli Zaretskii has issued what amounts to a last call for new features to appear in Emacs 27.1. He’s planning to create an Emacs-27 release branch in about a week. After that, only bug fixes for the release will be accepted.

There’s still a lot of testing and release candidates to be done, of course, but we’re getting close.

I’ve long been an admirer of Chris Wellons’ work and have written about him many times. I almost always learn something new and useful from his blog posts so of course I take notice when he publishes something new. His latest post is about building a linear congruential pseudo-random number generator. If you know anything at all about linear congruential PRNGs you might wonder what there is to say. Quite a bit, it turns out. Although they are very simple—just the recurrence relation \(x_{n+1} = x_{n} × C + A \bmod M\) where \(x_{0}\) is the seed—properly choosing the values of \(C\), \(A\), and \(M\) is important.

A good strategy is to choose prime numbers with certain properties for \(A\) and \(C\) and Wellons shows how easy this is to do using the excellent Calc utility built into Emacs. One of the things I learned from the post was that Calc has a function for the Miller–Rabin primality test. That’s not on the Calc Cheat Sheet so I’d long since forgotten about it and always cranked up my own Scheme version when I needed to check for the primality of large numbers. Having it available in Calc is a big win, especially since I’m probably making the calculations in Calc anyway.

Like all of Wellons’ posts this one is interesting and instructive and definitely worth your time.