Parse the results of buckets()

[ryantxu]

When trying to parse the output of the query buckets() I get a failure on:

Here is the csv:

#group,false,false,false,false,true,false,false
#datatype,string,long,string,string,string,string,long
#default,_result,,,,,,
,result,table,name,id,organizationID,retentionPolicy,retentionPeriod
,,0,grafana,059b46a59abab001,059b46a59abab000,,604800000000000
,,0,_tasks,059b46a59abab002,059b46a59abab000,,259200000000000
,,0,_monitoring,059b46a59abab003,059b46a59abab000,,604800000000000

[ryantxu]

ahh. Seems the header order is out of sync.... if I make the headers: datatype,group,default it works

#datatype,string,long,string,string,string,string,long
#group,false,false,false,false,true,false,false
#default,_result,,,,,,

Unclear where/how this should be fixed... either:

• server sends them in the same order
• client accepts them in either order

[vlastahajek]

It was once discussed in #110. Yes, the QueryTableResult parser expects the #datatype annotation as the first row and the Query method takes care of this.
How do you get the csv?

[jsternberg]

The flux CSV is meant to be parsed with the annotations in any order and we won't guarantee the order so the parser should be capable of parsing them in any order.

For helping with this specific issue right now, the annotations that are in the CSV will be returned in the same order as the entries in dialect. This seems to be constructed correctly for DefaultDialect where the annotations are specified in the specific order that works with the client. As a temporary solution, you might be able to change your client to request the annotations in the same order.

[ryantxu]

After digging a bit more, we are not hitting this problem with a real environment -- just when saving files and loading them for tests. In this case, we can manually reorder the headers and get the results needed

[vlastahajek]

Just to let you know, that custom order of annotations is now supported since #134. Only the datatype annotation is required.