« Day 16 | Day 18 »

(TBH, it was a really rough week for me, unrelated to the changelog. So my plans to work on it for 4-5 hours/day on the weekend weren’t implemented; now is 20:00 Sunday evening, and I’ll probably do something to move forward, yet not by any means close to what I was planning.)

So, today’s just moving forward with a standard library part. My rule for a changelog to follow what’s in the NEWS.md file, which meant that the standard library section becomes smaller and smaller each year: with extraction of its parts to gems (split into default ones and bundled ones), the NEWS.md mostly stops tracking those gems changes, telling to

See GitHub releases like Logger or changelog for details of the default gems or bundled gems.

…which is a) not really convenient (to check through all the standard gems you rely upon to see whether they’ve added some new features), and b) sometimes impossible (the standard gems not maintaining a detailed changelog), and I want at some point to tackle this problem too, and start re-adding a centralized list of the notable changes to the main language changelog… But probably not this year, and definitely not today.

Even if the current surface to cover is small, it is not without peculiarities and challenges. Say, a passing note in NEWS.md:

  • ext/readline is retired

…calls to:

  • Investigate the related discussion (and some other ones);
  • Decide on amount of code and explanations necessary, and, like, write them;
  • Notice that Reline (to which now Readline docs are redirecting) has a significant lack of documentation, compared to the “old” Readline and make a TODO (probably post-release? we’ll) see to try fix that;

Or, I investigate a new argument chars: to Random.alphanumeric, and is surprised it expects an array of characters (and not a string of them… or, if it wants to be generic, not an Enumerable). I investigate the change and find out it was just following the underlying private API which powered #alphanumeric. I can’t say I find the new API fully justified, but just thinking about raising the question about rarely needed API, a week before the release, with a proposal to complicate the (1-line) implementation— well, probably everybody can use their time better.

Or, investigating the Socket#recv and related methods return value change, I stumble upon the fact that change is implemented not in the class mentioned in the news (BasicSocket, not Socket), and that at least one of the methods has the contradictory documentation now.

And I am still not sure what does “Name resolution such as Socket.getaddrinfo, Socket.getnameinfo, Addrinfo.getaddrinfo, etc. can now be interrupted.” means (even after reading the ticket), but I am not really sure it deserves much digging.

Anyway, with that, the whole yearly changelog, with very, very rough explanations and examples is in place. Woohooo! Commit: 1c77be7

« Day 16 | Day 18 »