Strongtalk vs Exupery etc
Locked
 
|
Hi folks!
(personal reflection follows) There has been a lot of talk recently on this list regarding Strongtalk and how we could possibly benefit by that work in the Squeak community. Personally, having tracked Bryce's work on Exupery since he started, I think Exupery is *much* more interesting and especially much more realistic: - It is written *in* Squeak. This is a biggie. - It is written *for* Squeak. Bryce is one of *us*, it is being designed and tailored for Squeak use in Squeak projects by Squeakers. Don't underestimate this factor. - It works *today* in Squeak (even if we want it to be even faster). Not vapour. - Since it is an "offline" compiler it can go even further (than Strongtalk or a JIT) performance wise. - It integrates and interoperates with the current VM. This is probably crucial for large scale adoption in the community. Then we also have Ian's work on Pepsi/idst etc which is very interesting, but in a much longer time perspective. That work is also being done with Squeak "in the back of his head" - which of course makes it much more likely to benefit the Squeak community if it gets completed. It could then offer us an alternative lower level platform (replacing the VM essentially). So while I also think Strongtalk is "cool stuff" (but it has been around for ages) I actually don't think it will end up having much practical "impact" on Squeak. But I am happily proven wrong. My money (for a jump in low level performance) is on Exupery and Idst/Pepsi - and in general I think we should be starting to take Exupery into our long term planning since it is now becoming a real practical option. Sure, send speed is still not super duper - but when it goes up I bet people will start paying real attention. regards, Göran |
|
|
I don't see any of this in terms of either/or. While it is true that some
effort may be diverted by working with Strongtalk to develop a Squeak VM it doesn't mean that it will derail all other efforts. The question is should we as a community work with Strontalk to help make a Strongtalk VM for squeak possible. Even if we do nothing else but make Strongtalk better, does that really hurt smalltalk in general or Squeak specifically? It appears that Strongtalk will have its own community and will move forward with or without support from Sun, and with or without support from Squeak. Since the work will proceed anyway, and since strongtalk is already going to benefit from contributions from the squeak community, shouldn't we benefit from that work? I would think that we should not limit our options. The opportunity we have now is clear, work to support a Strongtalk vm version for squeak, and we could have a close relationship with another open source community, or pass on this opportunity, lose some people anyway, and further fragment the general open source Smalltalk community. Jecel may be right and it is not possible because of competing self interest to get the community to think of the benefit of the community as a whole, but I would think that now is the time to try. Decisions like this are difficult, if people feel that their efforts are being minimized. We really need to stress that the work that has been done on the squeak VM, and Exupery is extremely valuable and will continue to be very important. We can help Strongtalk and benefit the general Smalltalk community at the same time by creating a bridge between Squeak and Strongtalk for those in the community that might benefit from it. Ron Teitelbaum > From: [hidden email] > Sent: Tuesday, September 19, 2006 6:37 AM > > Hi folks! > > (personal reflection follows) > > There has been a lot of talk recently on this list regarding Strongtalk > and how we could possibly benefit by that work in the Squeak community. > Personally, having tracked Bryce's work on Exupery since he started, I > think Exupery is *much* more interesting and especially much more > realistic: > > - It is written *in* Squeak. This is a biggie. > > - It is written *for* Squeak. Bryce is one of *us*, it is being designed > and tailored for Squeak use in Squeak projects by Squeakers. Don't > underestimate this factor. > > - It works *today* in Squeak (even if we want it to be even faster). Not > vapour. > > - Since it is an "offline" compiler it can go even further (than > Strongtalk or a JIT) performance wise. > > - It integrates and interoperates with the current VM. This is probably > crucial for large scale adoption in the community. > > Then we also have Ian's work on Pepsi/idst etc which is very > interesting, but in a much longer time perspective. That work is also > being done with Squeak "in the back of his head" - which of course makes > it much more likely to benefit the Squeak community if it gets > completed. It could then offer us an alternative lower level platform > (replacing the VM essentially). > > So while I also think Strongtalk is "cool stuff" (but it has been around > for ages) I actually don't think it will end up having much practical > "impact" on Squeak. But I am happily proven wrong. > > My money (for a jump in low level performance) is on Exupery and > Idst/Pepsi - and in general I think we should be starting to take > Exupery into our long term planning since it is now becoming a real > practical option. Sure, send speed is still not super duper - but when > it goes up I bet people will start paying real attention. > > regards, Göran |
|
|
In reply to this post by Göran Krampe
> My money (for a jump in low level performance) is on Exupery and
> Idst/Pepsi - and in general I think we should be starting to take > Exupery into our long term planning since it is now becoming a real > practical option. Sure, send speed is still not super duper - but when > it goes up I bet people will start paying real attention. > > regards, Göran > +1 |
|
|
In reply to this post by Ron Teitelbaum
Hi!
"Ron Teitelbaum" <[hidden email]> wrote: > I don't see any of this in terms of either/or. While it is true that some I don't either. :) > effort may be diverted by working with Strongtalk to develop a Squeak VM it > doesn't mean that it will derail all other efforts. I actually did not mean that either. :) In fact - I just wanted to present my 2 öre about why Exupery probably is much more interesting (to me as a Squeaker) than Strongtalk, despite the current "buzz" about Strongtalk on squeak-dev. I think Exupery is underestimated in general. But I am an OSS fan and to each his own etc - just keep scratching all those itches. Remember that in the end it doesn't really matter much what "we decide" - because that is not how it works. The only thing that matters is what each individual developer decides to do. I would hardly think any of us will dig into Strongtalk code just because someone said on squeak-dev that it would be "good" for Squeak. :) And the same goes for any other effort of course. In short - have fun, whichever way you like - and things will work out for the best. :) regards, Göran |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |