Missing async utils

[Exidex]

Utilities that do exist in x11rb but not in x11rb-async, list possibly not exhaustive

• x11rb::properties::*
• x11rb::atom_manager
• x11rb::wrapper::*
• x11rb::image::*
• x11rb::extension_manager::*

Another very minor difference is that cookie module is private and reexported in async, but public in non-async version, so x11rb::cookie::Cookie works but x11rb_async::cookie::Cookie doesn't (x11rb_async::Cookie works)

I personally, currently, only use atom_manager, WmClass and WmHints so that would be sufficient, but given that x11 is dying I would understand not adding any of those

[psychon]

Quick first reaction:

• This covers all the hand written code, I think. So far, x11rb-async has mostly the equivalent of the generated code for x11rb (and the parts needed to make that work). These utilities would need to be copied and reimplemented. :-(
• No one will miss extension_manager. It is a utility for implementing your own connection class.

I personally, currently, only use atom_manager, WmClass and WmHints

The two properties classes can at least be half reused. You would have to sent the GetProperty request yourself and get the reply, but then at least the from_reply() functions can be used. This is not very ergonomic, but at least covers half of the use case.

but given that x11 is dying

It's been dying ever since I learnt what X11 is. I do not think it will really die any time soon.

Besides that: I do not have anything useful to say about this besides "I will not have time to work on any of this any time soon" (and of course the "I would hope to be able to do this without copy&paste, but... 🤷 )

[Exidex]

I have async copy-paste-adapt versions of atom_manager and WmClass, here and here respectively. Porting is was fairly straightforward.

I would hope to be able to do this without copy&paste

Not sure atm but if I would have motivation to contribute, would this mean that you would prefer to have current properties rewritten into a generator so that "duplicate" is generated for both cases or copy-adapt-would be ok?

[psychon]

would this mean that you would prefer to have current properties rewritten into a generator so that "duplicate" is generated for both cases or copy-adapt-would be ok?

Dunno. I would prefer if this were possible, but I fear that this code becomes much more complicated if done that way.

Random quick idea for the WmClass code would be: How much would really be needed? One would just need async versions of WmClass::get() and a new WmClassCookie type, while WmClass::from_reply() could be just reused (since x11rb and x11rb-async both use the same GetPropertyReply type from x11rb-protocol). So... perhaps this is semi-doable.

if I would have motivation to contribute

If you have some motivation, feel free to just cook something up and send a PR. I do not know what is nicely possible, so the person trying to figure that out and making a proposal wins over the person on the side line who does not help (= me).