Add SELECT DISTINCT queries support #264
Conversation
Codecov Report
@@ Coverage Diff @@
## master #264 +/- ##
==========================================
+ Coverage 61.32% 61.53% +0.20%
==========================================
Files 68 71 +3
Lines 6345 6574 +229
==========================================
+ Hits 3891 4045 +154
- Misses 1938 1996 +58
- Partials 516 533 +17
Continue to review full report at Codecov.
|
sql/planner/hash_set.go
Outdated
| func (s documentHashSet) generateKey(d document.Document) (uint64, error) { | ||
| defer s.hash.Reset() | ||
|
|
||
| err := d.Iterate(func(field string, value document.Value) error { | ||
| var buf []byte | ||
| buf, err := key.AppendValue(buf, value) | ||
| if err != nil { | ||
| return err | ||
| } | ||
|
|
||
| _, err = s.hash.Write(buf) | ||
| if err != nil { | ||
| return err | ||
| } | ||
|
|
||
| buf = buf[0:] | ||
| return nil | ||
| }) | ||
| if err != nil { | ||
| return 0, err | ||
| } | ||
|
|
||
| return s.hash.Sum64(), nil | ||
| } |
There was a problem hiding this comment.
This function relies on Iterate function implementation, so if it does not guarantee order of field iteration (for example map-based) then hash will be incorrect.
For now, Genji does not have such implementation, but it can be added later.
I think document.Value/document.Document should have some Hash function to use it in hash table.
There was a problem hiding this comment.
We should not rely on the implicit behavior of Document implementations as the Document interface doesn't guarantee the order of iteration.
Instead, we must use the document.Fields() function, which gets the document keys and sorts them, and use GetByField for each key.
There was a problem hiding this comment.
pk()/literal tests were failing, when I used documentMask.GetByField, it tries to get a value from masked document, but literals and functions values are not stored in document.
So, I added IterateInOrder function.
There was a problem hiding this comment.
It appears to be a bug, they should be stored in DocumentMask:
genji> create table foo;
genji> insert into foo(a) values (1), (2), (3);
genji> select pk() from foo;
{
"pk()": 1
}
{
"pk()": 2
}
{
"pk()": 3
}
|
@tdakkota thanks! Can you please mark the discussions as resolved? It looks like only PR author can do that in GitHub… |
tdakkota
force-pushed
the
feature/select-distinct
branch
from
c44d95d
to
0a8c1a2
Compare
Adds new
DISTINCTkeywordUses hash set to make projection result unique.