Compilation error when using group by about sqlpp11-connector-postgresql

[matthijs]

For postgresql you need to specifiy all columns in the group by statement (except for the aggregate), so it should look like this:

auto records = _connection(
    select(
        reports.id, 
        reports.aoi, 
        reports.toi, 
        reports.roi, 
        reports.story, 
        reports.created, 
        reports.properties,
        reports.author,
        users.firstName, 
        users.lastName,
        count(places.id)
    ).from(
        reports.join(users).on(reports.author == users.id).left_outer_join(places).on(reports.id == places.report)
    ).group_by(reports.id, reports.aoi, reports.toi, reports.roi, reports.story, reports.created, reports.properties, reports.author, users.firstName, users.Lastname)
     .limit(limit).offset(offset).order_by(reports.id.asc()).unconditionally()
);

[neel]

But in PgAdmin SQL query works only with group by reports.id.

I tried including all the fields. the reports.properties is of type json but in sqlpp11 it is mapped as text

It gave me exceptions

ERROR:  could not identify an equality operator for type json
LINE 1: ...rts.toi,reports.roi,reports.story,reports.created,reports.pr...
                                                             ^
unhandled exception ERROR:  could not identify an equality operator for type json
LINE 1: ...rts.toi,reports.roi,reports.story,reports.created,reports.pr...
                                                             ^

[rbock]

Interesting. I thought that @matthijs is right, but maybe there are rules for joins that allow this? I´d be thankful if you could dig up the rules that allow to use group by reports.id only.

However, the rules for select in the library are as stated.

Since you apparently cannot group by the JSON field, I see two options:

1. Use verbatim. That is a bit hacky, but probably the fastest way.
2. Restructure your query to join with a select. Something like this (taken from test/Select.cpp):

  auto rowCount = select(t.alpha, count(t.alpha)).from(t).unconditionally().group_by(t.alpha).as(sqlpp::alias::a);
  db(select(f.omega, rowCount.count)
         .from(f.join(rowCount).on(f.omega == rowCount.alpha))
         .unconditionally()
         .order_by(rowCount.count.asc()));