Yes but... Knative is supposed to define some "standards" for bulding, serving 
and eventing.
Also I do not look after any "decent" UI, building in OpenWhisk is (and must 
be) a hidden detail.

You send the sources and run the image. ActionLoop images are already able to 
precompile. The difference would be that at init time I "build" basically 
creating an image from a runtime with an additional layer with the code of the 
action (compiled in case of go swift and rust) that can be automatically 
executed, with no delay of /init (images will "autoinit" with that code). I 
already implemented that in ActionLoop.

This thing as I described is small enough that can be implemented in a 
reasonable amount of time, and it is fully "Knative" compliant. If I do not do 
it an "openwhisk on top of knative" I am wasting my time as the current 
OpenWhisk is already way better.

I want just to create a "MVP" and from there understand if merging, replace or 
reuse the pieces. Unfortunately only when you have something that exists and 
works you are able to understand what to do next properly.

-- 
  Michele Sciabarra
  mich...@sciabarra.com

----- Original message -----
From: Martin Henke <martin.he...@web.de>
To: dev@openwhisk.apache.org
Subject: Re: A plan to (re) implement OpenWhisk on top of Knative
Date: Monday, May 20, 2019 2:47 PM

Bertrand,

I am not directly involved in Knative Build or Tekton so I might be wrong on my 
dead end sentence.

For what I got ,I think It was realised that a build pipeline on Kube and the 
task to build docker containers from a given source
is a general problem to solve and in most ways independent from Knative.

The good side on this that step is that Jenkins X (the next version of Jenkins) 
is embracing Tekton already as a runtime.
So some would probably get a decent Build UI.

Regards,
Martin


> On 20. May 2019, at 14:17, Bertrand Delacretaz <bdelacre...@apache.org> wrote:
> 
> On Mon, May 20, 2019 at 2:07 PM Martin Henke <martin.he...@web.de> wrote:
>> ...As Knative Build seems be on a dead end...
> 
> Wow, already? That stuff seems to be competing with JavaScript
> frameworks in terms of short lifetimes these days ;-)
> 
> -Bertrand

Reply via email to