uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Klügl <peter.klu...@averbis.com>
Subject Re: FeaturePath vs. FeatureValuePath
Date Mon, 28 Sep 2015 08:36:42 GMT
There is of course also the option that I extend FeaturePath with the
missing functionality. However, to be realistic, we need a solution
before we can upgrade our dependency to a stable UIMA release that would
contain the changes.



Am 27.09.2015 um 15:40 schrieb Peter Klügl:
> Hi,
> yes, we have a current use case. Right now we are thinking about
> either using the deprecated api or implementing it ourself (only the
> parts we need currently). Will the deprectated stuff will be removed
> with the v3 release or can be count on it that can be used for a
> longer time? ... well it has been deprecated for some time and should
> then maybe be removed.
> Best,
> Peter
> Am 24.09.2015 um 23:44 schrieb Marshall Schor:
>> Hi,
>> I'm not familiar with this area.  I did some looking, and found the
>> original
>> note which motivated this change
>> http://mail-archives.apache.org/mod_mbox/incubator-uima-dev/200801.mbox/%3C4794BA4A.7030002@michael-baessler.de%3E
>> (Did a history on FeatureValuePath, found the UIMA-718 ref to
>> deprecate the
>> "old" way, and it had this link).
>> This stuff probably hasn't been worked on since 2008-ish.
>> Is there an actual use case for wanting feature arrays in the middle?
>> In v3, it might work to replace all this with Java 8 stream + lambdas
>> :-) as a
>> more general, standardized approach.
>> -Marshall
>> On 9/23/2015 9:40 AM, Peter Klügl wrote:
>>> Hi,
>>> there are two interfaces and classes to access nested feature values
>>> FeaturePath(Impl) and FeatureValuePath(Impl). The later has been
>>> deprecated in favour of the former, but it looks like there is
>>> functionality getting lost in this transition. With FeatureValuePath it
>>> is possible to have feature arrays in the middle of the path whereas
>>> FeaturePath only allows these at the very end of the path.
>>> Is this initial investigation correct? What is the current status of
>>> this functionality?
>>> Best,
>>> Peter

View raw message