Using the cygwin version to find packages in a separate mingw installation 🙃

Until I can be bothered to work it out that is

OK if you want this cursed setup you can use WSL , you’re already in abusive relationship with your OS

Trying to get the opam github across to do unholy things with pkgconf on Windows in the vain hope it might compile ocgtk

Got to keep them on their toes (and why are they allowed to be culturally ignorant?)

starting a group chat with “mates” doesn’t quite hit the same

Oh this seems to be the iOS app that is buggy- I was able to recover my session on Claude Code Web 🙃

It would be nice to be able to clear context with it too, and be able to pick a working directory instead of having it anchored at startup (to be able to switch projects easily) or even just be able to spawn new sessions remotely.

I do like the Claude Remote Control feature but its still so unreliable. It seems to get stuck at least a couple of times a day such that l have to access the session directly and continue it (permissions prompts seem to be where it is most buggy)

Buy Single Serve Pour Over Coffee Pouches Take your coffee experience to the next level with our Single Serve Pour Over Coffee Pouches! Discover 12g of Dipacci Coffee, get yours now!

I was so happy to have these when travelling in Japan, so I’m glad to see someone is making them in Australia too dipacci.com.au/products/sin...

Alas, not quite enough. There are a whole heap of symbols that don’t document when they were introduced. I think Rust’s bindings have this in their override files, so I might be able to use them for reference

I'm super close to finishing a Next.js-like experience for OCaml on top of Dream: file-system routing, server components, typed routes, and a CLI that hides most of the dune machinery

with LSP, error overlay, hot reload, melange integration...

Other platforms like freebsd and mac wont build because the glib version seems to be pulling in symbols from libpcre that don’t exist (it’s like an older system version of pcre is installed but glib has been compiled against a newer homebrew version, but it’s not installed)

The issue is that GIR cannot represent version information on enum and bitfield members - this information is only available in doc comments. So now I’m having to build a manual override system and extract all the doc comments and pray this is enough 😅

Opam CI tests the code against a large number of Linux distributions, and many of the old ones carry older versions of Gtk and related libraries that don’t compile with code generated with new definitions, even though I have version #if .. #endif guards in the C code

So I’ve put out a preview release of ocgtk (my Ocaml bindings for gtk4), but an opam release will take a fair bit more work 😭

Oh that macos is just an experimental build. I can drop the version ranges from conf-gtk4 and hopefully get this thing building on all the linuxes (that matter)

I just got versioned API guards to work in ocgtk but it won’t build on macos in opam CI/CD still due to some excluded API parts 😭😭

I’ve learned how to make Tesco style pitas and now they’ve become currency within the extended family 🤦‍♂️

This milestone is done. Just need to cleanup the repo, write a roadmap, and cut a release

At the same time, the weeds are extremely overgrown, so there’s more work to do 🤦‍♂️

Sydney is going through that midautumn period where the weather is perfectly moderate (warm days cool nights) and there is just enough rain for the garden to look lush.

Something that can’t think or feel will never have good taste

No wonder us software developers are loving it for code: precise, but unmoored (specific when we don’t care about requirements)

Destruction of art, destruction of meaning

Thank you, I hate myself already

oh no now I’m writing my own testing framework

Folks this Is how it starts

I’m trying to create tests for ocgtk that load a Gtk UI and interacts with it, checking that actions and their results render.

There doesn’t seem to be anything specific for this on the Linux desktop, but it appears I can get by somewhat with the ATSPI (accessibility API) that GNOME publishes.

there are other small and large milestones for ocgtk to get something substantially useable with gtk and with sufficient api coverage:

Glib and Gobject types support
Signal handling
Function and constant generation
Record field accessors
API docs natively generated to odoc format
Callback handling