Is Spy of Self 4.0 written in Self? If not (probably not...), can it or similar utility be written in Self and how slow will it be? Can it be written in Self without making the system heavily loaded?
-- just a whimsical question of a Self lover who is interested in the relationship between program execution monitoring and reflection
a*
Arata Kubota writes:
Is Spy of Self 4.0 written in Self? If not (probably not...), can it or similar utility be written in Self and how slow will it be? Can it be written in Self without making the system heavily loaded?
Speed isn't so much the problem -- the main issue is that the spy needs to run asynchronously with the VM. For example, there would be no way to update the display during a garbage collection if the spy was written in Self itself.
-Urs
Speed isn't so much the problem -- the main issue is that the spy needs to run asynchronously with the VM. For example, there would be no way to update the display during a garbage collection if the spy was written in Self itself.
-Urs
I see. Isn't a multiple VMs system (or multi-threaded) possible? a*
Arata Kubota =?ISO-2022-JP?B?GyRCNVdKXUVEGyhCIBskQj83GyhC?= wrote:
Is Spy of Self 4.0 written in Self?
No, it is written in C++.
If not (probably not...), can it or similar utility be written in Self and how slow will it be?
I don't think that speed would be the problem, but rather that the "reflective calls" included in the virtual machine don't give enough details for this application.
Can it be written in Self without making the system heavily loaded?
One problem of having this program load the system is that it would distort the results we are interested in - the observer changes the system being observed. That would be worse than any slowdown in the other applications, in my opinion.
-- just a whimsical question of a Self lover who is interested in the relationship between program execution monitoring and reflection
The problem is exactly the presence of some "holes" in Self 4.0's reflection facilities. Nothing that would be hard to fix, though.
- Jecel
From: Jecel Assumpcao Jr jecel@lsi.usp.br
Thank you for the guidance.
One problem of having this program load the system is that it would distort the results we are interested in -
the observer changes the system being observed
because of the observer and observed run on the identical VM? What was the reason for choosing the name "mirror"?
a*
self-interest@lists.selflanguage.org