On 08/08/2013 08:18 PM, pablo pazos wrote:
> Hi Bert,
>
> I have the need to clarify some of your assertions about my previous 
> messages:
>
> Pablo advises you to use a relational database.
>
> That?s not true. I supposed David were working with relational, but he 
> didn't confirmed that.

OK, no hard feelings about that. I must have misunderstood. I apologize.

>
> openEHR persistence should be DBMS independent, so I'll never advise 
> to use relational over other thing without knowing requirements. *DBMS 
> option should depend on requirements. (e.g. how data will be used / 
> consumed).*

One of the requirements is that it must be able to run path based queries.

> *
> *
> Also, there are different level of persistence needed: for local 
> systems, shared/federated systems, mobile and cloud based. For local 
> systems and some shared systems, I would recommend relational. Maybe 
> also for mobile persistence on the device. For other kinds, I'll 
> suggest XML/JSON based DB. And for some applications, I would 
> recommend EAV or path-value.
> I know some of those are different from your solution, but that is not 
> mean that are not suitable for a huge space of solutions. I don't 
> believe in one-fits-all solutions.

I did not advise a solution to David, I just said that a relational 
database could be good for starting, but would not be suitable for a 
full featured OpenEHR database.
I recommended against a relational database.

>
> IMO, this doesn't give David the answer he needs. Is good to give him 
> options. Consider he's on a learning process.
>
> I don't think that is suitable for a good working kernel,
>
> I'm not talking about a kernel, I'm talking about persistence. One 
> architect can put that layer on a kernel or as a service on the cloud, 
> depends also on requirements.
>
> because you cannot run path-based queries against it, but for a start 
> it might work.

Service, cloud, in the end, it must go to a disk, that is the part I am 
talking about.
You can leave this decision to others, cloud, architect, 
service-providers, but somewhere this decision must be made.

>
> That's not true. Anyone can run path based queries against any type of 
> DBMS, relational included. You just need a query transformer as 
> recommended by AQL articles.

Yes of course it is possible, but when talking about relational, it 
needs a layer which transforms from AQL to a Codd optimized database.
In fact, it will be a transformation from AQL to SQL into a RDBMS schema.
I don't think it is even possible for a few programmers to write that.
So that solution is for me out of the question.

In your case, as you state it, you want the cloud to solve it, or 
somewhere else where the architect (whoever that may be) puts it. That 
is fine.

I want to solve it myself, with the people I work with.
I think that is where the money is. In the solving of the technical 
problems. That is what I am discussing.
There is no need to interfere in that discussion if you don't want to 
regard that as your problem.

But if you do want to regard it as a situation you want to handle, then 
I am very pleased, if you explain/discuss how you are planning do it.

Thanks in advance
Bert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<http://lists.openehr.org/pipermail/openehr-technical_lists.openehr.org/attachments/20130808/d03ecb82/attachment.html>

Reply via email to