flink-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephan Ewen <ewenstep...@gmail.com>
Subject Re: I dont understand
Date Mon, 13 Nov 2017 19:52:20 GMT
A lot of projects evolve out of use cases, Flink evolved out of the Big
Data ecosystem, which was a few years ago dominated by Hadoop.

The next version of Flink is also independent from Hadoop, but bundles it
as an optional dependency for the convenience of many users.


On Nov 12, 2017 02:02, "Cristian Lorenzetto" <cristian.lorenzetto@gmail.com>
wrote:

> Sincerely i dont understand how it is developed this project.
> You are essentially developing a streaming management permits complex
> internal states, permits to split streams in parallel way , to merge and
> reduce streams in statefull way. It is great idea. But this is a generic
> concept. Does this abstract concept   need hadoop for exiting ?No. It is a
> generic and abstract solution for many problems. Does this concept need to
> know the client-server concept for working  and need to be part of a client
> -server concept for being what is ?No
>
> You developed flink project embedding strict dependencies with hadoop, with
> yarn in a monolitic project that pratically blocks anyone to use it in a
> different way from how you embedded inside your solution. You merged a
> abstract algorithmic idea with many other things not strictly related. In
> addition you developed it as server solution.You merged a generic algorithm
> with a concrete application of this algorithm. This method to develop
> blocks anyone want use your abstract concept for doing something over. The
> world evolves when a tecnology is easy to reuse for creating other more
> complex solution over, when the people neednt to go inside what it is
> invented or delivered for a new  idea ,because they can compose your idea
> with others. This composition process permits to evolves the tecnologies.
> In short , the right way is to develop short libraries containing the
> strict logic necesssary for implementing that idea. And then you can
> implement you solution/application based on it.
> Think for a moment to another big solution how is lucene. It is developed
> as a indipendent library. In a decades , tons of open sources and
> commercial solutions were developed over lucene. The lucene team didnt
> developed it like a server. Anyway then  they developped elasticsearch
> server. If they would have developed directly elastic search do you think
> the open source history was the same? Your abstract idea is very good and
> can be aplied in many different ways . Do not block the future.To rewrite
> all for a new application based on the same idea requires a effort and
> money few people /companies have. Sincerelly a so grave error for a open
> source foundation like Apache foundation is incomprehensible. I hope it
> could be solved in future for the good of open source future.
>
> All the best for the open source world
>
> bye
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message