Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

WIP: Experiments and hacks attempting to implement SwiftUI-like layout flow #379

Draft
wants to merge 25 commits into
base: main
Choose a base branch
from

Conversation

mortenbekditlevsen
Copy link
Contributor

Please don't look at this yet - it's horrible! :-)

Currently I made a lot of types in TokamakCore public in order to perform the layout in the renderer. The layout should actually reside in TokamakCore with only a slim api to the renderer.

The GTK renderer is my current experimentation playground.
Instead of always returning a GtkWidget for the AnyWidget conformance it's now ok to be a 'virtual' (pure layout) widget and not return any actual control.
A 'context' is used to keep a stack of offsets used during layout - as well as a reference to the last actual widget container where all child elements should be added.
The only implementation of the sizing/layout is for Text where the GtkWidget is actually measured to return it's size, as well as _ShapeView whose size doesn't rely on any actual measuring.
Box (a helper type used by HStack and VStack in the GTK renderer) implements an extremely naive version of a layout. It's completely wrong, but shows how the sizing/layout could be tied together.
For the actual algorithm I propose to use the MIT licensed NotSwiftUI sample code from Objc.io. Perhaps only paying members of the Swift Talk video series have access, but I do have access through my work.
This naturally requires Tokamak to include the LICENSE text with copyright notice from objc.io.

I am currently experimenting with Overlay, but I think now is the time to move the size and layout protocol requirements into TokamakCore.

Apologies for the extremely rough/messy state of this draft PR. :-)

@MaxDesiatov MaxDesiatov added the GTK renderer Extending GTK support label Feb 13, 2021
…nto TokamakCore to make everything fit nicely together. Use CGPoint instead of CGAffineTransform for context since the latter has issues in .concatenate. Will create a separate PR to fix that.
…nitialized array used by CGAffineTransform concatenation
…a bit of an annoying protocol conformance to Target
…this logic to StackHelper to keep it separate and to reuse between HStack and VStack
@mortenbekditlevsen
Copy link
Contributor Author

Hi @j-f1 ,
Thanks so much for your work on this PR.
I've unfortunately not had much time to put into this lately - due to the usual: work, family, life.

The scope of the task also became slightly daunting - there are a few general areas that I've not tried to tackle yet at all:

  1. Performing the layout on every state change instead of just initially
  2. Dealing with how 'native' components could be interleaved with the layout. For instance the children of a button in the gtk version ought to be wrapped in a 'fixed' widget where the context from that point should then be relative to top-left corner of the button.
  3. Figuring out how measurements might even be performed in a browser. I've seen suggestions that adds hidden versions of elements to the dom in order to perform measurements, but that seems like it might be a wasteful way to go.
  4. Probably more that I can't think of right now...

I did try to merge main into this branch earlier, but I had the idea that the layout functionality might live on PrimitiveView, but after messing about with this concept I figured that this was not the right idea since that required all PrimitiveViews to be able to perform layout in the TokamakCore module, whereas for some views this work must be done in each of the renderers. And this means that the layout must be 'dynamic' (in the sense that protocol conformance must be tested at runtime before calling into the layout code).

I haven't yet looked into all of your additions, but please let me know if there is something you would like to discuss. :-)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
GTK renderer Extending GTK support
Development

Successfully merging this pull request may close these issues.

3 participants