Global Smalltalk Image
Locked
 
Administrator
|
Is this a good idea? Aik-Siong Koh
Global Smalltalk Image Contains everyone's code. Ideas are never lost. Easy to search everything. (Indexed for fast search). Remove duplications. (Automated as much as possible). Includes all forks all the time. Code growth will slow with size as reuse becomes more and more effective. Programming becomes more and more high level with high reuse of lower levels. Lower levels becomes more and more optimized, stable and tested. Automated runtime packager to make small application images. |
|
|
On Sun, 10 Oct 2010, askoh wrote:
> > Is this a good idea? Aik-Siong Koh > > Global Smalltalk Image > Contains everyone's code. > Ideas are never lost. > Easy to search everything. (Indexed for fast search). > Remove duplications. (Automated as much as possible). > Includes all forks all the time. > Code growth will slow with size as reuse becomes more and more effective. > Programming becomes more and more high level with high reuse of lower > levels. > Lower levels becomes more and more optimized, stable and tested. > Automated runtime packager to make small application images. An image with all previous code cannot exist, there are conflicting implementations of a lot of stuff. The code isn't in the image by default. A database with full history would be useful. IIRC Marcus Denker had this idea a few years ago. Here are some problems to solve: - one has to find all previous code (it's almost impossible, but the sources/changes files can be found) - we don't know the license of all methods - the more forks are involved the harder to make it through - some forks can have the same method under different method name/method category/class/class category Levente > > > -- > View this message in context: http://forum.world.st/Global-Smalltalk-Image-tp2970621p2970621.html > Sent from the Squeak - Dev mailing list archive at Nabble.com. > > |
|
Administrator
|
In reply to this post by askoh
We can start with working images for Squeak, Pharo, Cobalt, etc, and with applications like Seaside, Etoys, etc. Each can be put in their own namespaces for starter. Once they are in, a search utility will bring all the common code to a 'Common' namespace and all the duplicated code in the other namespaces removed. Individual developers can then use more and more of the 'Common' code in their projects and borrow code from other namespaces. The automatic search utility will move code to Common if the code is use several times in different namespaces. I think the different projects will turn out to be quite thin as time goes on.
The Global Smalltalk Image can be archive weekly so that nothing is ever lost. Everyone has the benefit of the entire knowledge of the Smalltalk community forever. We can develop tools to traverse the image and its history so that reinventing should not happen. Only truly new ideas will be coded and hopefully at a very high level. All the best, Aik-Siong Koh
|
|
|
In reply to this post by askoh
On 10/10/10 1:39 PM, askoh wrote:
> Is this a good idea? Aik-Siong Koh > > Global Smalltalk Image > Contains everyone's code. > Ideas are never lost. > Easy to search everything. (Indexed for fast search). > Remove duplications. (Automated as much as possible). > Includes all forks all the time. > Code growth will slow with size as reuse becomes more and more effective. > Programming becomes more and more high level with high reuse of lower > levels. > Lower levels becomes more and more optimized, stable and tested. > Automated runtime packager to make small application images. > > http://forum.world.st/My-own-Squeak-direction-tt621647.html#a622131 Lawson |
|
|
In reply to this post by Levente Uzonyi-2
On 10/10/10 1:56 PM, Levente Uzonyi wrote:
> On Sun, 10 Oct 2010, askoh wrote: > >> >> Is this a good idea? Aik-Siong Koh >> >> Global Smalltalk Image >> Contains everyone's code. >> Ideas are never lost. >> Easy to search everything. (Indexed for fast search). >> Remove duplications. (Automated as much as possible). >> Includes all forks all the time. >> Code growth will slow with size as reuse becomes more and more >> effective. >> Programming becomes more and more high level with high reuse of lower >> levels. >> Lower levels becomes more and more optimized, stable and tested. >> Automated runtime packager to make small application images. > > An image with all previous code cannot exist, there are conflicting > implementations of a lot of stuff. The code isn't in the image by > default. > A database with full history would be useful. IIRC Marcus Denker had > this idea a few years ago. > > Here are some problems to solve: > - one has to find all previous code (it's almost impossible, but the > sources/changes files can be found) > - we don't know the license of all methods > - the more forks are involved the harder to make it through > - some forks can have the same method under different method > name/method category/class/class category > > > Levente > I mentioned before. Lawson |
|
|
In reply to this post by askoh
Hi Askoh
I think it is a good idea though not easy to implement --- or probably only to a certain extend. A Global Image has a lot of code to learn from (positive and negative....). You give many more good reasons. I assume you might not have watched the efforts of Edgar closely where he tried to load many projects into FunSqueak (based on trunk). Just search the mailing list for the last two months. A recent image is here. http://ftp.squeak.org/Experiments/FunSqueakCog4.2-10577-alpha.9.zip Edgar has quite a number of interesting things in there but as well quite a number of incompatibility problems have surfaced during this process. As others have pointed out there will be incompatibility problems but the fact is that these problems will only surface if people actually try to build a "Global Smalltalk Image"....... So, please go ahead and build one which suits your interests and share it with us..... There is more room in http://ftp.squeak.org/Experiments (It will probably not be "Global" but the aim is going into that direction). HTH --Hannes On 10/10/10, askoh <[hidden email]> wrote: > > Is this a good idea? Aik-Siong Koh > > Global Smalltalk Image > Contains everyone's code. > Ideas are never lost. > Easy to search everything. (Indexed for fast search). > Remove duplications. (Automated as much as possible). > Includes all forks all the time. > Code growth will slow with size as reuse becomes more and more effective. > Programming becomes more and more high level with high reuse of lower > levels. > Lower levels becomes more and more optimized, stable and tested. > Automated runtime packager to make small application images. > > > -- > View this message in context: > http://forum.world.st/Global-Smalltalk-Image-tp2970621p2970621.html > Sent from the Squeak - Dev mailing list archive at Nabble.com. > > |
Re: Global Smalltalk Image
|
|
In reply to this post by LawsonEnglish
Lawson English wrote on Sun, 10 Oct 2010 16:43:18 -0700
> See http://wiki.squeak.org/squeak/5637 and related links and the thread > I mentioned before. A "single world wide image" was indeed my motivation in sketching out Neo Smalltalk Modules. This was inspired by my observations of a group of students trying to share a single image using the Kansas interface in Self. There were serious problems, but the advantages were so compelling that I can't imagine a way forward that isn't fixing the problems by introducing a more robust infrastructure. In a very similar direction, see the latest VPRI paper: http://www.vpri.org/pdf/tr2010001_worlds.pdf "Worlds: Controlling the Scope of Side Effects" Alessandro Warth, Yoshiki Ohshima, Ted Kaehler, Alan Kay -- Jecel |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |