Simon McVittie

For context, I am the maintainer of dbus-python, or possibly the unmaintainer. This mostly involves rejecting changes that would make it lose its single most important feature, namely "is compatible...

> Looks like the DBus has standarized on '\0'-terminated binary strings. Do you know any APIs that provide non-'\0'-terminated "ay"s? Yes: every byte array that is not "basically a string",...

If you want to coordinate among binding and library authors, please (encourage them all to) contact [email protected]. > Implementing higher-level, per-interface wrappers should not be necessary. I can understand the...

> I still think that there is no reason to send \0-terminated strings over the wire Whether you consider the reasons to be sufficiently good or not, you're too late...

> Has a bug already been opened on GDbus project ? If it was, it would be closed NOTABUG. This is not a bug in GDBus. It is a convention...

> I've looked at GDBus, and it looks like they're copying the values while converting between messages and GVariants Indeed, GDBus isn't zero-copy in this way (but GVariant is; and...

The Flatpak developers would also find it useful if the loader searched the [XDG basedirs](https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html), similar to how Vulkan ICDs are located.

> why not revert e0eed78 altogether? (Because, it seems the CheckBusNameExists method is only used here.) In the OO languages I'm most familiar with (C# is not one of them),...

> That protected method was introduced in the commit to revert. Yes, and that's why reverting the commit would be an ABI break.

I don't think e0eed78 made things better; what it seems to be aiming for can never be fully reliable, and I think the 0.7 API was a better way to...