uima-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Osborne, John D" <josbo...@uabmc.edu>
Subject RE: [External Sender] Re: shall we start a UIMA blog?
Date Tue, 24 Apr 2018 20:12:36 GMT
Thanks Marshall, I think I have misunderstood the V3 change.

I do have JCas classes corresponding to UIMA Types. For example a document class that includes
various metadata I am interested in and write to a database. I had the impression that V3
might remove the requirement to have a type system, thereby allowing arbitrary java objects
to be added to the CAS. By arbitrary I mean (for example) a java POJO not defined in the type
system.

The use case I was thinking of was using directly hibernate annotated POJOs in the CAS instead
of mapping between them and my type system defined objects.

 -John


________________________________________
From: Marshall Schor [msa@schor.com]
Sent: Tuesday, April 24, 2018 10:03 AM
To: dev@uima.apache.org
Subject: Re: [External Sender] Re: shall we start a UIMA blog?

I'll try and post some info about this on some section of the website.

The type system for UIMA has an "optional" connection to Java objects.
>From your query, should I assume you've created JCas classes corresponding to
(some of) the UIMA types?

In v2, each UIMA type that had equivalent JCas class, had two classes, named
following a naming convention where the UIMA type was the JCas class name, for
one, and the other was the same with a suffix of "_Type".
In v3, the 2nd class (with the name ending in _Type) is no longer used or needed.

The details of the JCas implementation changes between v2 and v3, but the
user-facing APIs are the same.

Can you say more about what kind of best practices you're looking for?  Do you
have a simple example to share? That will help guide the discussion.

-Marshall


On 4/24/2018 9:46 AM, Osborne, John D wrote:
> I'd like to see a blog or any other information about v3 - specifically best practices
around navigating the relationship/migration between the (old?) type system and java objects,
the latter of which may already have an ORM to a relational database.
>
> ________________________________________
> From: Marshall Schor [msa@schor.com]
> Sent: Tuesday, April 24, 2018 7:35 AM
> To: dev@uima.apache.org
> Subject: [External Sender] Re: shall we start a UIMA blog?
>
> right you are.... I've had that exact experience.
>
> And I like your idea about tweeting.
>
> Thanks. -Marshall
>
>
> On 4/21/2018 3:50 PM, Richard Eckart de Castilho wrote:
>> IMHO blogs tend to be started, then a small number of posts are published, and
>> then they are abandoned. But if you feel you really wish to maintain a periodical,
>> no objections.
>>
>> You could also tweet from time to time and point people to specific new parts
>> of the UIMA documentation or to the migration document? You could write a
>> new beginners tutorial as part of the UIMA documentation and tweet about that.
>>
>> Cheers,
>>
>> -- Richard
>>
>>> On 19.04.2018, at 18:56, Marshall Schor <msa@schor.com> wrote:
>>>
>>> According to https://urldefense.proofpoint.com/v2/url?u=http-3A__apache.org_dev_project-2Dblogs&d=DwIDaQ&c=o3PTkfaYAd6-No7SurnLtwPssd47t-De9Do23lQNz7U&r=SEpLmXf_P21h_X0qEQSssKMDDEOsGxxYoSxofi_ZbFo&m=BUA5eDvlAci3g4ufVMEBJJ4RUzBU5ET_wX4O6OwEFkY&s=n8eXCLgduPqLCKOYqWa3-Uefe_UJfubMnJTD4eJixy8&e=
projects can have a blog, just
>>> by asking.
>>>
>>> I'm wanting to occasionally blog about uima version 3 to increase awareness.
>>>
>>> We could use it for other topics too...
>>>
>>> WDYT?
>>>
>>> -Marshall
>


Mime
View raw message