I really would love to and I will probably do so to take my mind of the refactoring I am engaged in as I broke my SecureSession code. Note this basically includes a small outside header describing the contents of a FEC encoded msg. There will be an inside header coming later. I am not yet ASN1 encoding it. The picture I used to have, if you can imagine it is a 9 layer OSI stack, where layer 8 is the marketplace of running or invokable applications and the 9th, the meta, describes the work. One job where we did this we pushed event structure definitions and writer and reader interfaces for event flows. So all of this definition is stuffed into the meta, accessible anywhere in the distributed app. Furthermore, we injected decoders into the presentation layer and injected running readers and writers into the control layer (part of layer 8... 8 1/2) I hope this gives you a bit of a better picture what I am speaking about. When my world stabilizes I will try to get back to you soon, but I cannot promise you I will remember - my apologies for that. regards, robert On 12/16/2015 05:22 PM, phil@highoctane.be wrote:
Can you make a more detailed vision (with a picture) about this?
As I have a cluster running and am paid to do some stuff with all you list, I can use some time exploring those avenues.
Phil
On Wed, Dec 16, 2015 at 7:07 PM, Robert Withers <robert.w.withers@gmail.com <mailto:robert.w.withers@gmail.com>> wrote:
Excuse me:
With interfaces I think groovy. With object serialization I think avro. With header encoding I think ASN1DER with dynamic extension, we need variable encoding of statistical objects. With packages I think gradle.
this would be powerful, robert
On 12/16/2015 01:00 PM, Robert Withers wrote:
I'd like to add that this is a real challenge to existing players in BigData, all of who offer solutions in this exact space. Whoever controls it's specification, controls it's cloud market.
My back of the envelope proposal (my apologies for my abrasiveness with this approach) is such:
We develop a dynamic cloud meta & control solution: replicated, eventually consistent and so on. We define the serialization, interfaces and package identification to this meta as open-source standards.
With interfaces I think Avro.
With serialization I think ASN1DER encoding, plus non-fixed dynamic structure extension.
With package identification I think Gradle. Do we have a groovy compiler?
I thought I would round out my thinking as I believe this to be an important opportunity. Given that the new internet is going to be governed by always running, maximally interactive execution sites, it will naturally move to Smalltalk, so I do not think there is anything for anyone to be concerned about.
Patience and not reacting emotionally to misinformation in the marketplace will show true colors. Business is warfare. We have the advantage.
peace, robert
On 12/16/2015 12:38 PM, Robert Withers wrote:
Allow me to speak a little on my ideas of a distributed meta layer and control layer in cloud deployments with Squeak/Pharo, hopefully in bounds.
A distributed system must have an metadata repsoitory describing the traffic and activity. If you look at deployment, code that works with a type and version of data must be present. With packaging and dynamic loading, chunks of code registered to a data type/version in the meta demand loads a achunk to do directed work in a particular deployed image. BigData event flows and data analysis like to get replicated and relocated, in realtime. Supporting this in the meta definitional layer and the 1/2 control layer starts to do dynamic, late-binding in the network. This is the advantage Squeak/Pharo image-based environments bring to the business table.
Thank you and apologies, robert
On 12/16/2015 09:57 AM, H. Hirzel wrote:
Is there a Pharo implementation? https://avro.apache.org/docs/1.2.0/
On 12/16/15, Robert Withers <robert.w.withers@gmail.com <mailto:robert.w.withers@gmail.com>> wrote:
Please consider Avro.
robert
On 12/16/2015 08:58 AM, H. Hirzel wrote:
If you want to move data to Java then you probably go for JSON or a particuar XML format.
On 12/16/15, H. Hirzel <hannes.hirzel@gmail.com <mailto:hannes.hirzel@gmail.com>> wrote:
No, it is a Smalltalk format.
On 12/16/15, Denis Kudriashov <dionisiydk@gmail.com <mailto:dionisiydk@gmail.com>> wrote:
2015-12-16 14:41 GMT+01:00 H. Hirzel <hannes.hirzel@gmail.com <mailto:hannes.hirzel@gmail.com>>:
It probably should be noted here as well that
http://pharo.gemtalksystems.com/book/PharoTools/SIXX/
Just to know does java implementation exists?
-- . .. .. ^,^ best, robert
-- . .. .. ^,^ best, robert
-- . .. .. ^,^ best, robert