Browser Example Broken

[abhillman]

Observed:

Expected:

Notes:

trunk gives the above "observed" screenshot. The "expected" screenshot is given with 0.4.0-beta2 and this diff (else compilation fails):

diff --git a/examples/browser/toolbar.rs b/examples/browser/toolbar.rs
index b4416dd..f3bcf68 100644
--- a/examples/browser/toolbar.rs
+++ b/examples/browser/toolbar.rs
@@ -35,12 +35,12 @@ impl BrowserToolbar {
         let back_button = Button::new("Back");
         let mut back_item = ToolbarItem::new(BACK_BUTTON);
         back_item.set_button(back_button);
-        back_item.set_action(|| Action::Back.dispatch());
+        back_item.set_action(|_| Action::Back.dispatch());
 
         let forwards_button = Button::new("Forwards");
         let mut forwards_item = ToolbarItem::new(FWDS_BUTTON);
         forwards_item.set_button(forwards_button);
-        forwards_item.set_action(|| Action::Forwards.dispatch());
+        forwards_item.set_action(|_| Action::Forwards.dispatch());
 
         let url_bar = TextField::with(URLBar);
         let url_bar_item = ToolbarItem::new(URL_BAR);

#30 appears to be the commit that breaks the browser example.

[ryanmcgrath]

Hmmm, are you saying that the webview isn't loading anything, or that back doesn't work?

And what version of macOS are you testing against...?

[abhillman]

Hmmm, are you saying that the webview isn't loading anything, or that back doesn't work?

The webview is not loading anything aginst trunk, but does load against 0.4.0-beta2.

And what version of macOS are you testing against...?

$ sw_vers
ProductName:		macOS
ProductVersion:		14.5 # Sonoma 14.5
BuildVersion:		23F79

$ uname -a
Darwin ok.local 23.5.0 Darwin Kernel Version 23.5.0: Wed May  1 20:17:33 PDT 2024; root:xnu-10063.121.3~5/RELEASE_ARM64_T6031 arm64

[ryanmcgrath]

Hmmm, might be down to a difference in obj/objc2. @madsmtm might know more, otherwise I'd have to find time to dig in - and I'll be busy + traveling the next two weeks.

If you wanted to tinker and debug, a webview not loading sounds like something not getting retained somewhere - either the webview itself, or the webview's delegate. It's where I'd start looking at least.

[abhillman]

[edit: submitted before seeing your last message @ryanmcgrath]

Glad to take a deeper dive in case you have any rough ideas and would enjoy any support.

I have taken a cursory look at the changes to webview between trunk and 0.4.0-beta2 https://github.com/ryanmcgrath/cacao/compare/0.4.0-beta2..trunk, but nothing major jumps out.

[abhillman]

If you wanted to tinker and debug, a webview not loading sounds like something not getting retained somewhere - either the webview itself, or the webview's delegate. It's where I'd start looking at least.

Agreed regarding retainment. I'm not sure if meaningful, but I did see some differences in the webview configuration after objc2 was introduced; i.e. a "strong pointer" against 0.4.0-beta2 but something different against trunk:

It's not clear to me without more studying if this is meaningful.

[abhillman]

Indeed perhaps @madsmtm may have some thoughts...

[madsmtm]

The old StrongPtr is the same as the objc2::rc::Id, there's no difference there. The issue is indeed very likely that something that was previously (erroneously) leaked is now no longer getting retained long enough.

If you could put your code up somewhere, then it'd be easier to diagnose? Otherwise, I'd suggest to try sprinkle std::mem::forget(obj) in various parts of your code.