[ 
https://issues.apache.org/jira/browse/DRILL-7092?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Arina Ielchiieva updated DRILL-7092:
------------------------------------
    Description: 
Drill maps concept is more close to Struct concept. Good write up by Paul about 
this:
https://github.com/paul-rogers/drill/wiki/Drill-Maps. Internally such objects 
are represented by MapVector which in future will be renamed to StructVector 
(similar was done in Arrow project). Drill also does not support true maps, 
which in future will be also implemented with proper naming (MapVector).

Nevertheless, internal map / struct presentation does not affect users plus 
before schema provisioning they did not deal with this concept at all.
Initially map complex type in Schema Provisioning schema was represented with 
map keyword followed by angle brackets: col1 map<m1 int, m2 int>. Given second 
thought to map notion, looks like it's better to use struct in schema 
definition and internally convert it to map metadata (again which eventually 
will be renamed to struct). Such approach would ease map to struct renaming 
(since it will be done internally) and users will be using struct naming as 
before. Once Drill will implement true map, schema provisioning will allow map 
type for schema.


  was:
Drill maps concept is more close to Struct concept. Good write up by Paul about 
this:
https://github.com/paul-rogers/drill/wiki/Drill-Maps. Internally such objects 
are represented by MapVector which in future will be renamed to StructVector 
(similar was done in Arrow project). Drill also does not support true maps, 
which in future will be also implemented with proper naming (MapVector).

Nevertheless, internal map / struct presentation does not affect users plus 
before schema provisioning they did not deal with this concept at all.
Initially map complex type in Schema Provisioning schema was represented with 
map keyword followed by angle brackets: col1 map<m1 int, m2 int>. Given second 
though to map notion, looks like it's better to use struct in schema definition 
and internally convert it to map metadata (again which eventually will be 
renamed to struct). Such approach would ease map to struct renaming (since it 
will be done internally) and users will be using struct naming as before. Once 
Drill will implement true map, schema provisioning will allow map type for 
schema.



> Rename map to struct in schema definition
> -----------------------------------------
>
>                 Key: DRILL-7092
>                 URL: https://issues.apache.org/jira/browse/DRILL-7092
>             Project: Apache Drill
>          Issue Type: Sub-task
>            Reporter: Arina Ielchiieva
>            Assignee: Arina Ielchiieva
>            Priority: Major
>             Fix For: 1.16.0
>
>
> Drill maps concept is more close to Struct concept. Good write up by Paul 
> about this:
> https://github.com/paul-rogers/drill/wiki/Drill-Maps. Internally such objects 
> are represented by MapVector which in future will be renamed to StructVector 
> (similar was done in Arrow project). Drill also does not support true maps, 
> which in future will be also implemented with proper naming (MapVector).
> Nevertheless, internal map / struct presentation does not affect users plus 
> before schema provisioning they did not deal with this concept at all.
> Initially map complex type in Schema Provisioning schema was represented with 
> map keyword followed by angle brackets: col1 map<m1 int, m2 int>. Given 
> second thought to map notion, looks like it's better to use struct in schema 
> definition and internally convert it to map metadata (again which eventually 
> will be renamed to struct). Such approach would ease map to struct renaming 
> (since it will be done internally) and users will be using struct naming as 
> before. Once Drill will implement true map, schema provisioning will allow 
> map type for schema.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to