Video Link: Jayaprabhakar Kadarkarai - Create your own DSLs This is technically a link to a talk contained within a video. The entire video is worth watching it’s the lightning talks at the first Software Should Work conference. In this particular lightning talk, Jayaprabhakar Kadarkarai explains why he thinks you should be more willing to create your own DSLs. He gives an example of one for creating a bounded set of HTML. ...
Video Link: Reliability lessons from SQLite at Software Should Work Richard Hipp the creator of SQLite, clearly one of the best pieces of software in the world, gives a talk on reliability lessons from SQLite. This is very good talk that is well worth watching. There is a certain kind of time-warped feel to this talk, Richard discusses several testing strategies that I think have been at least in the standard vocabularly for two to three decades, if not always standard practice. He sometimes seems to kind of refuse to use the standard vocabulary for these things, I wonder if he kind of independently discovered these things himself. Again though, it’s well worth watching, the reliability techniques he points out are things most seasoned software developers would already know about, but still he is essentially saying “these are the most important things for SQLite”. There are hundreds of relability techniques that software developers already know, picking out the most important ones for a piece of software as successful as SQLite is a great way to prioritise what you already know. ...
Link: Twenty years of Pandoc An interesting look back at the first twenty years of Pandoc by its creator and maintainer John MacFarlane. If you’re not interested in the history of improvements and evolution of Pandoc, you may still find the discussion on the choice of Haskell interesting. You can read the Prehistory section for why Haskell was chosen and the Retrospective: the choice of Haskell section for a discussion of how well that has gone. ...
The following post 23 languages, one I can check showed up on my feed. The author talks about using gettext for internationalisation (i18n) and how they used an LLM to translate into 23 languages other than English, with only one of those 23 (German) being one the author could personally check. If you look into translation engines, such as gettext which the author is using, and ICU message format, you will find that there are a lot of differences between languages and dealing with these differences is tough. A particularly thorny problem is choosing the correct form of a sentence depending on the number of some noun. In English, we have it pretty easy, it’s usually one for 1, and another for all other numbers. If zero is possible, that’s often a special case as well, but it’s also usually special cased in the actual functionality as well. For example, suppose you have a webstore, on a past order, you may have a button that allows the user to add all the items from that order to the current cart. In English there will be three cases ...
Video Link: Dependency Cultures: Software Should Work An interesting talk by Richard Feldman well-known to most Elm developers and now working on his own Roc programming language. He starts with a very interesting observation, looking at the web pages for various programming languages and their dependencies. He makes the point that these websites are all doing a very similar job, although some have a built-in compiler/interpreter for the language so that visitors can try out the language directly on the website. This does change the website considerably from a simple static information page to an interactive app, or usually more a static information page that contains an interactive app. ...
Link: Functional Programming from First Principles: Part 1 motivation An interesting start to what looks like a series of articles on functional programming. I’ve often found that once people give functional programming a real and proper try they get it quite quickly, but that many programmers resist it. Perhaps they give it a half-hearted try but it’s almost as if they want to dislike it. There may be some selection effects going on here, so perhaps there are people that give it a real and solid try but still give-up as it’s not to their tastes, but I never hear of them because they’re not in my circles. ...
Link: A testing conundrum A thoughtful blog post by Ned Batchelder regarding a difficulty in testing a class that on the face of it should be straightforward to test. The class seems straightforward to test because: This is a pure function: inputs map to outputs with no side-effects or other interactions. It should be very testable. The code is basically a hashing function, for which he wants “equal” values to hash to the same hash, and non-equal values to hash to different hashes. ...