fi, en, (sv, ja, hu, yi) | avatar by https://twitter.com/udonkimuchikaki


libera.chat, irc.sortix.org
nortti
microblog (that is, a blog with small entries)
microblog.ahti.space/nortti

hikari-no-yume
@hikari-no-yume

a few people were interested in a technical write-up. the thing's too damn big and complicated to explain in just a single post, but here's a write-up of one particular aspect of it! https://hikari.noyu.me/blog/2023-04-13-touchhle-in-depth-1-function-calls.html


You must log in to comment.

in reply to @hikari-no-yume's post:

Oh, I'd love to write one, but there's just so many moving parts that it'd be hard to actually cover the whole thing. I actually wanted to write a proper blog post about it for the release, but I realised that my high standards for blog posts meant I'd never get it done. In the end I wrote a twitter thread which actually worked pretty well when turned into blog form, but the way I thought about it was pretty different…

I can see myself eventually writing about it by reducing the scope a bit. For instance, I could write about what pieces are involved in making a simple Objective-C method call work.

super focused stuff would be cool! and honestly the more little self-contained write ups the better (maybe easier to write and then we get "more" stuff :p) fun bugs or cool implementation tricks would be cool to hear about too. but i like 150% get the effort and perfection angle

hello this rules and i love the use of <ruby> tags

btw my rss aggregator initially refused to load your hybrid blog page/atom feed, and i think it's because it's served with content-type application/xhtml+xml. i told it to ignore that and it loaded fine, but i worry that other people might not be able to subscribe because of this

thanks :3

yeah, I know it's the wrong MIME type for an Atom feed, it's a workaround for Safari refusing to attempt to display the page if I do claim it's Atom, because Safari used to have an RSS/Atom feed reader, and now it displays a message telling you it can't read such feeds, even though I made sure the page can be displayed as styled XML.

maybe I need to send a different MIME type depending on the Accept or User-Agent headers or something

in reply to @hikari-no-yume's post:

Since there's one stub function for each dynamically-linked function, the dynamic linker can infer which function it needs to do linking for by inspecting the call stack to see which stub function it was called from.

As far as I understand from the snippet, the call stack does not contain any information about the stub used. I'd presume the information is gotten from the r12-register instead, as it points to the function pointer