Use "GROUP BY" twice

[stdedos]

I have this query:

```dataview
TASK
WHERE !completed
GROUP BY meta(section).subpath
```

Which gives me a very good estimate of what I'm looking for.

I'd like to now also GROUP BY path (or, hopefully, its auto-linking counterpart).

Separately, both work. However, together I either get:

• **-** (3) as the first group, for

  ```dataview
  TASK
  WHERE !completed
  GROUP BY meta(section).subpath
  GROUP BY path
  ```

• or

  Dataview: Every row during operation 'group' failed with an error; first 1:
  
                  - No implementation of 'meta' found for arguments: null
  

  for

  ```dataview
  TASK
  WHERE !completed
  GROUP BY path
  GROUP BY meta(section).subpath
  ```

Any ideas? Any clarification I missed from the docs?

[holroy]

Whenever you do a GROUP BY two things happens, unique variants of the group expression is stored into key, and the list of all the other fields and values are stored into rows.<fieldname>. So in theory you'd need to do GROUP BY path and GROUP BY meta(rows.section).subpath or similar, but that doesn't really make sense since you're then grouping on a list of values.

So what kind of output would you expect from the double grouping in this case?

---

A simple double grouping which seems to work:

```dataview
TASK WHERE !completed
LIMIT 40
GROUP BY meta(section).subpath
GROUP BY key[0]
```

This first groups by the section subpath, and then on the first character of that key. Is it usefult? Maybe not, but it does show a double grouping.

[stdedos]

So what kind of output would you expect from the double grouping in this case?

I'm trying to generate something like that:

• path/to/file1
  • link > to > header 1
    • Task A
    • Task B
• path/to/file2
  • link > to > header 2
    • Task C
    • Task D
  • link > to > header 3
    • Task E