Today work has focused on two things; Working through the last few PRs I had pending feedback on in rust-lang/libc, and looking into the Idris programming language.
I started off by looking into the last two PRs and issue I had to comment on. The patchset introducing the time64 cfg I had almost finished working through yesterday, and today I rounded it off with a quick fix to CI.
The rust-lang/rust PR implementing the MCP for embedding the -current release supported by OpenBSD targets has also been addressed. This one only needed a quick little fix in new documentation.
The reviewer for that PR also mentioned that we could temporarily add to the set of expected cfgs for diagnostics in the compiler the new value introduced by the target_env in OpenBSD. I previously noted this was a subpar solution.
A good implementation goes through extending the whole cfg system for this type of ranged “version” values. There is an RFC for this, but it is still a long way off; Hardcoding values i something I would prefer to avoid.
I also commented on the tracking issue a rust-lang/libc maintainer recently opened up for time_t matters. I was pinged because of my work on this area, and added commentary on how Qualcomm’s Hexagon and Haiku handle this.
In terms of side-project work, yesterday I mentioned I would be looking into Idris as a replacement for my choice of ML languages. The project I had attempted tackling in the last few weeks was fine; I just wanted purity.
Idris offered this and dependent typing on top of that. From what I have read of the documentation, the direct manipulation of types in what ML languages would consider type equations is beautiful. It feels even better than using GADTs.
I have thus far only been going through the installation of Idris’ most popular package and toolchain manager. I gave up on system package management through homebrew a few days ago, so I have had to build a few libraries from source.
For now, I have had great success in compiling GMP and Chez Scheme. Prior requirements for GNU coreutils seemed unnecessary on my system, so I hope to avoid installing those.
I also decided to profit this chance and install some other libraries and tools I needed for recovering some speed in my workflow ever since I stopped using a number of tools.
None.
The rust-lang/libc work is done. I can now choose to either follow up on the siginfo_t PR with another patchset that removes the skips in the build script, or continue working on the ctest extension for modules. We shall see.
In terms of side-project work, I expect to have a fully functioning Idris compiler by early tomorrow. With that, I should be capable of working through the documentation’s crash course and the community tutorial.