[Development] Proposal for "container-oriented deterministic memory manager"

[Phil Bouchard]

On 01/05/2017 03:09 PM, André Pönitz wrote:
> On Thu, Jan 05, 2017 at 07:57:54AM -0600, Thiago Macieira wrote:
>> Em quinta-feira, 5 de janeiro de 2017, às 07:26:52 CST, Phil Bouchard
>> escreveu:
>>>> AFAIU QtQuickCompiler has nothing to do with memory management, its main
>>>> purpose is reduction of start up time and obfuscation of sources.
>>>
>>> Ok I assumed that execution time would be affected because the code is
>>> compiled.
>>
>> The code is compiled anyway. The difference is only *when* it is compiled: at
>> release time of your application or when your user launches it (JIT).
>
> It's not the only difference. A JIT compiler has typically not the same
> scope/abilities/optimization opportunities as a real compiler, not to mention
> deficiencies in a language that has 'double' as only numeric type.
>
> So "compiler" and "compiler" are apples and oranges in this case.

Right, g++ is much more portable than JIT.

>> JIT = Just Too Late (because it will start compiling the moment you need it)
>
> Correct.
>
> And done on each and every (often under-powered) device out there, if not even
> done on each and every application start, instand once when preparing binary
> packages.

In Brave New World we would have the QtQuickCompiler running on the fly, 
followed by g++ and the browser could "dlopen" the generated library.  I 
don't see any better solution than this.