Nenad Rakocevic
Nenad Rakocevic
This looks like some debug code I've used recently on master branch for investigating a GC-related issue, but that's not part of the code on github in `master` or `alert-on-error`...
So, I tried to add a dummy codec for Redbol code... there's a problem, it cannot work in the current way `load` works. `load` has two modes: a) When `/as`...
The only clean way would be to make all those refinements (and future ones) part of the codec API... though they were tailored for Red data, so they are not...
> A big one is how instancing should work The codec API just provides an `/options` refinement for passing a block/object of key/values pairs. How the codec is dealing with...
About refinements compatibility, we should have a standard way to define and check for that, both at mezz and R/S levels. I don't have a proposition for that right now....
I've pushed the alert messages for premature exit on errors. I won't modify `load` or codecs for now, until we get a better plan for that.
`return:` as a set-word is the most readable and human-friendly option we have. The core issue is elsewere: words collecting used by those functions is a form of static analysis...
It's not just GTK, same behavior on Windows.
@meijeru Did you forget to add a `Needs: 'View` entry in the header when compiling it? Compiling works fine here, and exhibits the same stack overflow error.
I can still get a stack overflow error on clicking on `Clear` button, though, it takes several manipulations, switching back and forth between inserting some texts and clicking the button.