Re: MV3 JShelter Chrome Release (Re: Yesterday meeting)

[Giorgio Maone]

I've just force-pushed (on both branches) a more testable version.

I'm sure there are bugs because the changes are huge, conceptually and 
the Chrome API themselves are still buggy and unstable.

I'd also like to make more explicit to users (maybe in the UI on 
startup?) that MV3at this moment requires JShelter (like any other 
extension based on dynamic script injections, like GreaseMonkey) to have 
users turn on the development mode toggle: see 
https://developer.chrome.com/docs/extensions/reference/api/userScripts#developer_mode_for_extensions_users

So expect me to push more changes later today and tomorrow.

 I've notice noticed that debug versions (compiled with "make debug") 
causes infinite recursion errors in the content wrappers on Chrome 
(where the debugging code calls wrapped APIs like performance.now() from 
within the wrapper).
Not sure whether it's a MV3-specific bug or it's already been happening 
in previous Chrome versions, but I'll get to it after any other more 
user-facing fix.

Thanks and best,
-- G

On 22/07/24 13:44, Libor Polčák wrote:
> > Saturday I've pushed both mv3 branches (JShelter and nscl) on their 
> > respective repositories.
>
> Hello Giorgio,
>
> I wanted to try before
>
> > However some more changes are needed to make Chrome's userScripts API 
> > work for us, which I plan to commit and push later today (afternoon 
> > or evening).
>
> However, I cannot build the extension:
>
> git submodule init
> git submodule update
> fatal: transport 'file' not allowed
> fatal: Fetched in submodule path 'nscl', but it did not contain 
> ac6785866c7cd2bb29654cbaa621531d35273131. Direct fetching of that 
> commit failed.
>
> It looks like JShelter refers to a non-existing commit in NSCL.
>
> Libor