While the story doesn’t focus on kernels (Linux) only, but also on DBMSs and such, perhaps it shows it is time for rethinking.
In the OS space, this rethinking has long been done, but until now been too much radical to gain a practical hold. The Hurd is perennially almost done, but it was conceived to solve precisely this problem: internalize some complexity in the architecture of the system, to simplify its interfaces, so that changes are better contained thus both simplifying development (and thus code review) and limiting side effects of new development (thus both simplifying code review, and making it less critical).
In the DBMS space, SQL is too complex, full of arbitrary limitations, inefficiencies, redundant-but-slightly-different ways of doing stuff and so on. It has been some years now the Third Manifesto proposed a radically simpler, more powerful class of D data languages, of which there are several prototypes and even a full-scale if proprietary and slightly deviant system.
Finally, the language problem. Until know being based on a functional language is considered an oddity, limiting the amount of talent available for Lisp (for instance) based projects. But given the simplicity of Lisp, and specially Scheme, one hopes Guile will become more widespread as an extension library, and Lisp as a systems programming language. Now we already have GNU Emacs, GNUCash, the Gimp, Sawfish; ideally all these would standardise on the Guile, and on the other hand GCL would become more widespread and fix its deficiencies.
By the way, the Hurd is but a stepping stone for a Lisp future.
Copyright © 2018, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds