[Pharo-project] Preparing my presentation of NativeBoost
Hi, i'd like to know what you would like to know about things you don't know about NativeBoost :) because i don't know what would you like to know more: - more in-depth about an implementation details - more about examples and how to use it just thinking, what i should focus on. For those who never heard about this project before, here the link to project home page: http://code.google.com/p/nativeboost/ See you in Edinburg. -- Best regards, Igor Stasenko AKA sig.
On Fri, Aug 5, 2011 at 1:00 PM, Igor Stasenko <siguctua@gmail.com> wrote:
Hi,
i'd like to know what you would like to know about things you don't know about NativeBoost :) because i don't know what would you like to know more:
- more in-depth about an implementation details - more about examples and how to use it
I much prefer this topic. Implementation details may be too low level and maybe few can follow you.
From a newbie point of view, I would be nice to make clear the relation between a JIT and NativeBoost. What they do, which are the differences, etc. For example, may NativeBoost help the JIT ? how? Then, I would like to hear how NativeBoost is related to FFI.
just thinking, what i should focus on.
For those who never heard about this project before, here the link to project home page: http://code.google.com/p/nativeboost/
See you in Edinburg.
-- Best regards, Igor Stasenko AKA sig.
_______________________________________________ Esug-list mailing list Esug-list@lists.esug.org http://lists.esug.org/mailman/listinfo/esug-list_lists.esug.org
-- Mariano http://marianopeck.wordpress.com
On Aug 5, 2011, at 1:04 11PM, Mariano Martinez Peck wrote:
On Fri, Aug 5, 2011 at 1:00 PM, Igor Stasenko <siguctua@gmail.com> wrote: Hi,
i'd like to know what you would like to know about things you don't know about NativeBoost :) because i don't know what would you like to know more:
- more in-depth about an implementation details - more about examples and how to use it
I much prefer this topic. Implementation details may be too low level and maybe few can follow you.
From a newbie point of view, I would be nice to make clear the relation between a JIT and NativeBoost. What they do, which are the differences, etc. For example, may NativeBoost help the JIT ? how? Then, I would like to hear how NativeBoost is related to FFI.
+1. I don't think you need more implementation details than a brief overview of how it works (store machine generated from smalltalk assembly code in methods with a special tag, execute that instead of method body if it's been generated). Then spend the rest of the time on how you use that, starting at the bottom and building some NB-versions of normal smalltalk methods directly, then how you've implement an FFI on top of it, and how it could fit into a JIT/other interesting stuff written entirely image-side. IE. focus on "how do/can I use it?" rather than "how does it work?" Cheers, Henry
Igor, i would like to know also the "why" i should use NB instead of other choices. Pros and cons. Looking forward to see your talk! Fernando On Fri, Aug 5, 2011 at 1:24 PM, Henrik Johansen <henrik.s.johansen@veloxit.no> wrote:
On Aug 5, 2011, at 1:04 11PM, Mariano Martinez Peck wrote:
On Fri, Aug 5, 2011 at 1:00 PM, Igor Stasenko <siguctua@gmail.com> wrote:
Hi,
i'd like to know what you would like to know about things you don't know about NativeBoost :) because i don't know what would you like to know more:
 - more in-depth about an implementation details  - more about examples and how to use it
I much prefer this topic. Implementation details may be too low level and maybe few can follow you.
From a newbie point of view, I would be nice to make clear the relation between a JIT and NativeBoost. What they do, which are the differences, etc. For example, may NativeBoost help the JIT ? how? Then, I would like to hear how NativeBoost is related to FFI.
+1. I don't think you need more implementation details than a brief overview of how it works (store machine generated from smalltalk assembly code in methods with a special tag, execute that instead of method body if it's been generated). Then spend the rest of the time on how you use that, starting at the bottom and building some NB-versions of normal smalltalk methods directly, then how you've implement an FFI on top of it, and how it could fit into a JIT/other interesting stuff written entirely image-side. IE. focus on "how do/can I use it?" rather than "how does it work?" Cheers, Henry
On 05.08.2011 18:02, Fernando Olivero wrote:
Igor, i would like to know also the "why" i should use NB instead of other choices. There are none? ;) If you mean "Why should I use an FFI built on top of NB over FFI/Alien?" that's an entirely different matter. NativeBoost (a smalltalk assembly DSL -> machinecode generator/executor) itself does not have any equivalents (that I am aware of at least), but the things you can build on top of it do.
Cheers, Henry
On 5 August 2011 18:55, Henrik Sperre Johansen <henrik.s.johansen@veloxit.no> wrote:
On 05.08.2011 18:02, Fernando Olivero wrote:
Igor, i would like to know also the "why" i should use NB instead of other choices.
There are none? ;) If you mean "Why should I use an FFI built on top of NB over FFI/Alien?" that's an entirely different matter. NativeBoost (a smalltalk assembly DSL -> machinecode generator/executor) itself does not have any equivalents (that I am aware of at least), but the things you can build on top of it do.
Yes, it is not yet-another FFI. It opens a lot more doors, and FFI is one of the closest ones, which already open :) -- Best regards, Igor Stasenko AKA sig.
+ 10 :) On Aug 5, 2011, at 1:24 PM, Henrik Johansen wrote:
I don't think you need more implementation details than a brief overview of how it works (store machine generated from smalltalk assembly code in methods with a special tag, execute that instead of method body if it's been generated).
Then spend the rest of the time on how you use that, starting at the bottom and building some NB-versions of normal smalltalk methods directly, then how you've implement an FFI on top of it, and how it could fit into a JIT/other interesting stuff written entirely image-side.
IE. focus on "how do/can I use it?" rather than "how does it work?"
participants (6)
-
Fernando Olivero -
Henrik Johansen -
Henrik Sperre Johansen
-
Igor Stasenko -
Mariano Martinez Peck -
Stéphane Ducasse