Invitation for technical discussion on next-generation Thunderbird

Magnus Melin mkmelin+mozilla at iki.fi
Sun Apr 23 18:27:54 UTC 2017


On 4/23/17 4:17 PM, Ben Bucksch wrote:
> Magnus Melin wrote on 23.04.2017 13:44:
>> While most things would be JavaScript, I doubt it's feasible to have 
>> a desktop application with no C++ at all.
>
> Aside from OS integration (e.g. "default email app" etc.), where do 
> you see the need for C++ code?

Depends on what level of application you're talking about, but I'm 
referring to an installable binary. What features you get from the "app 
shell"/platform depends on the platform, but obviously you're not going 
to be implementing things like encryption, or application update and 
similar things in JavaScript. These are vital things the platform need 
to cater for, but since it's bundled with the application I'd still 
count that very much as the application.

>
>> Without looking too much on that proposal, it strikes me that if it's 
>> a good idea, why isn't that standard?
>
> With that argument, we would never do anything new. Ever. Because if 
> it was a good idea, it would already have been done. So, per se, all 
> new ideas are bad.
>
No, it's clear you can experiment with that elsewhere, and see where 
that leads. Trying to shoehorn that experiment into the rewrite of mail 
components seems like an bad idea though. This is also a sure way of 
making adaption of ready 3:rd party libraries a non starter.


> If the collections were observable
>
> First, just because a collection is observable doesn't mean it's 
> writable by the consumer. These are distinct qualities.

Distinct sure, but the whole assumption that you'd observe the changes 
is quite moot when the collections are just used locally inside the 
component. The component probably do not want to observe data changes 
inside itself.

  -Magnus




More information about the tb-planning mailing list