Pharo 64bit planned?
Locked
 
|
I did not see any progress over the time in making a (stable) release
of pharo in 64 bit flavor. For servers, it is a common practice (and also a good one if no 32bit executables are planned) to completelly disable 32bit support (by not compiling any library in 32bit, disabling emulation of 32bit in kernel), and also performance, as under 64bit, 32bit code runs slightly slower. The above make enough reasons to make it interesting to have a 64bit version of pharo, but... Is it intended/in progress? Or nothing really at the moment? |
|
|
Hi David,
On Fri, Jul 17, 2015 at 3:22 PM, David <[hidden email]> wrote: I did not see any progress over the time in making a (stable) release 64-bit development of Spur is in progress. There is a somewhat functional 64-bit Squeak Spur image and a functional Linux 64-bit Stack interpreter. Coincidentally I'm currently working on an input event processing bug which only occurs in the real VM. You may know that the next release of Pharo (Pharo 6?) is intended to be a Spur release. Esteban Lorenzano is working on the Pharo Spur bootstrap. I am working in Squeak and my priorities are first, to get the Spur Squeak 64-bit image working fully on the Stack Interpreter and then to work on an x64 JIT VM. I expect that by the time Pharo 6 is ready to release, the 64-bit version will also be ready. Hope this helps. -- _,,,^..^,,,_ best, Eliot |
|
|
The idea is that we will do a Pharo4s in the fall that is just Pharo4+Spur. Pharo5 (next release) in spring 2016 will be based on spur of course, too. This has in addition lots image side improvements… e.g. - new text editor - the MetaLink model for behavioral reflection - breakpoints - First class Variables usable, but not yet used in the image - …. lots more Yes, we might then again do a Pharo5s that is Pharo5+64bit Spur… because Pharo6 will be a HUGE step: It will be based on Bloc, the replacement of Morphic that is now under active development. On top of Bloc there will be a complete, clean, modern set of widgets, a new text model and the next iteration of GT tools based on all that. Marcus |
|
|
In reply to this post by StormByte
> I did not see any progress over the time in making a (stable) release > of pharo in 64 bit flavor. This is that you do not look carefully enough. :) This is planned but it takes time as you can expect. - First spur new GC (32) that eliot designed to get working in 64 bits. - Second 64bits. Now if companies want to get it done faster, there is the consortium that is a non profit structure to collect money and hire more people. I can tell you if we would have a couple of millions euros invested into Pharo. Pharo would be massively different :) > > For servers, it is a common practice (and also a good one if no 32bit > executables are planned) to completelly disable 32bit support (by not > compiling any library in 32bit, disabling emulation of 32bit in > kernel), and also performance, as under 64bit, 32bit code runs slightly > slower. > > The above make enough reasons to make it interesting to have a 64bit > version of pharo, but... Is it intended/in progress? Or nothing really > at the moment? |
|
|
In reply to this post by Eliot Miranda-2
On Fri, Jul 17, 2015 at 4:31 PM, Eliot Miranda <[hidden email]> wrote:
and I've understood and fixed the event processing bug (which was nothing to do with 64-bits) so I can say that the Squeak Spur 64-bit Linux StackInterpreter is functional (again).
_,,,^..^,,,_ best, Eliot |
|
|
In reply to this post by StormByte
> On 18 Jul 2015, at 00:22, David <[hidden email]> wrote: > > I did not see any progress over the time in making a (stable) release > of pharo in 64 bit flavor. but there is, is just that things requires time :) - as already pointed, there is already a functional stackvm for 64 bits - pharo is being migrated to spur32 (which will happen next weeks) - right after, pharo will produce also 64bits versions to be tested with the stackvm - we will start work on FFI and Cog(JIT) for 64bits right after Esteban > > For servers, it is a common practice (and also a good one if no 32bit > executables are planned) to completelly disable 32bit support (by not > compiling any library in 32bit, disabling emulation of 32bit in > kernel), and also performance, as under 64bit, 32bit code runs slightly > slower. > > The above make enough reasons to make it interesting to have a 64bit > version of pharo, but... Is it intended/in progress? Or nothing really > at the moment? > > |
«
Return to Pharo Smalltalk Developers
|
1 view|%1 views
| Free forum by Nabble | Edit this page |