I have created a shared folder with Computer Science articles I read, understood and liked. At the moment, there are two items in there, but will be more over time. It’s not a surprise that both of them deal with the question of engineering a system as large as a programming language infrastructure and runtime and staying sane — this is a question that occupies lots of my thinking lately (I don’t know the answer yet, but promise not to withhold my findings). Here they are:

• P. Li, A. Tolmach, S. Marlow, and S. P. Jones, “Lightweight Concurrency Primitives for GHC”, in Proceedings of the ACM SIGPLAN workshop on Haskell, 2007;
• K. Wang, Y. Lin, S. M. Blackburn, M. Norrish, and A. L. Hosking, “Draining the Swamp: Micro Virtual Machines as Solid Foundation for Language Development”, in 1st Summit on Advances in Programming Languages (SNAPL 2015), 2015.

This does not mean, of course, that I only read two articles in my life. It’s just that these two I understand much better than the others and I see how they can immediately be useful in my programming practice. Understanding is a key point here: I remember that during my PhD time my approach to scientific articles was utterly wrong. Here are few rules I learned over time:

• Reading and understanding abstract is a prerequisite to even attempting to read an article. There’s a video of one of the Don Knuth’s lectures where he famously answers a question on performance of some tree-based algorithm from the audience and says something in the vein: “I remember seeing an article describing recent advances about this algorithm, but authors didn’t describe performance limits in the abstract, so not sure.” Abstract absolutely must contain each and every key important moment of the entire article.
• One consequence of the above is that there must be articles out there describing important things, but not doing well in their abstract and being completely overlooked because of that. My attitude towards that is sad acceptance: it is simply not possible to assume the best about every article and spend time reading it anyways when article’s abstract is not clear or not exhaustive.
• At the university I read articles trying to get an answer to what-questions: “what kind of new things people work on?” The better question, the way it seems to me now, is how-question: “how do people do those things they do?” The crucial difference between the two is that the second one implies at least a brief familiarity with the field for which the article is written. What-questions, I now think, are better answered on other media, like textbooks.