The feeds test cases are under asterix-app/src/test/resources/runtimets/queries/feeds
For the UDFs I believe the test cases are: asterix-app/src/test/resources/runtimets/queries/user-defined-functions I can prepare some simple examples till next week. -heri > On Oct 30, 2015, at 5:58 PM, Chen Li <[email protected]> wrote: > > @Heri: do you know where to find the test cases of AQL UDF, or have some > test cases of your own? Jianfeng can start from there. > > Chen > > On Fri, Oct 30, 2015 at 12:30 AM, Heri Ramampiaro <[email protected]> wrote: > >> Although I haven’t tested this (I have mostly used & created java-based >> UDFs), >> one can execute AQL calls from within a UDF. Feeds allows functions to >> execute arbitrary AQL statements (DDL, DMLs etc). I.e., I believe what >> you are trying to do is possible. >> >> For example you could do: >> (Given that you have a rec. type called Tweet, and a dataset >> ProcessedTweets >> >> create feed CoordTwitterFeed if not exists >> using “push_twitter" (("type-name"="Tweet”)) >> apply function find-intersection; >> >> (Here "find-intersection” is an AQL function that does the step specified >> under 2)). >> >> To do nr. 3 the easiest way is to have a modified version of the “hashTag” >> (let’s call this “tweetlocator”). Java-based UDF (see my previous message >> with the >> TweetLib example). You can then connect this as a secondary feed, >> connected to >> “CoordTwitterFeed”. >> >> E.g.: >> create secondary feed ProcessedTwitterFeed from feed CoordTwitterFeed >> apply function “tweelib#tweetlocator"; >> >> Thereafter you can connect the feeds to appropriate datasets. >> >> (Since the find-intersection is an AQL function, you can also call >> “tweelib#tweetlocator” >> inside this function). >> >> Best, >> -heri >> >>> On Oct 30, 2015, at 12:18 AM, Young-Seok Kim <[email protected]> wrote: >>> >>> I'm not sure whether the following UDF is possible or not, but hopefully >> it >>> is. >>> What we're trying to do is to have the following UDF. >>> >>> The UDF >>> 1) accepts an incoming tweet record through a feed job as an input and >> then >>> 2) takes a field, more specifically, coordinate field value from the >> tweet >>> record and sends a spatial-intersect query using the coordinate in order >> to >>> find out the corresponding county of the coordinates (we have created >>> AsterixDB instance which stores US county shapes records into a dataset >> and >>> created R-tree index on the polygon field of it, so the query will return >>> the county effectively using the R-tree index) >>> 3) creates a new tweet record consisting of the original record's fields >> + >>> the returned county value >>> 4) ingests to a tweet dataset in the AsterixDB instance. >>> >>> Can we have such an UDF? >>> >>> Best, >>> Young-Seok >>> >>> >>> On Thu, Oct 29, 2015 at 3:53 PM, Heri Ramampiaro <[email protected]> >> wrote: >>> >>>>> >>>>> 2. Can we use AQL function in those kind of feed UDFs? >>>> >>>> Can you give an example of what you are trying to do? >>>> I.e do you want to run an AQL inside a UDF or use an AQL function as a >>>> UDF connected to a running feed? >>>> >>>> -heri >> >>
