Question

Not sure why I am getting just one row with wrong count!

Following is the working code:

SELECT project_name, sub_project_name
FROM projects, sub_projects
WHERE projects.project_id = sub_projects.projects_project_id

Result:

project_name         sub_project_name
Bakken               Ghost fracture Study
Bakken               Bakken Mylo QAQC
Bossier              Doyle Boles K No.1
Eagleford            Kennedy Unit#1H
Eagleford            Wehmeyer Unit #1
Niobrara             Crow Valley 7-62-32-1M
Poland               Poland
Woodford             Ridenour Phase 2
Woodford             Teague 1-14H

Each sub_project has multiple tables and I am basically trying to see if a column in one of the tables has any row containing non-Null values and then show that count beside the sub_project in another column for each sub_project.

Here is what I get when I query the following statement:

SELECT projects.project_name, sub_projects.sub_project_name, COUNT( bl.bl_por ) AS porosity
FROM projects, sub_projects
LEFT JOIN bl ON sub_projects.sub_project_id = bl.sub_project_id
WHERE projects.project_id = sub_projects.projects_project_id

Result:

project_name         sub_project_name          porosity
Bakken               Ghost fracture Study      99

All the rows from every sub_project in the first row and nothing else.

Whats wrong here?

---

Edit:

Erwin’s solution nailed it. But I do not have sub_project_id in my tables. I just added it in one table to make it simpler for testing.

So I used Erwin’s GROUP BY suggestion to edit my statement and I get the right matrix shape but not the right counts. Counts are way off.

SELECT p.project_name, sp.sub_project_name, COUNT( bl.bl_por ) AS porosity
FROM projects p, sub_projects sp, wells w, cores c, samples s, inputs i
LEFT JOIN bl ON i.inputs_id = bl.inputs_inputs_id
WHERE p.project_id = sp.projects_project_id
AND s.sample_id = i.samples_sample_id
AND c.core_id = s.cores_core_id
AND sp.sub_project_id = c.sub_projects_has_wells_sub_projects_sub_project_id
GROUP BY p.project_name, sp.sub_project_name

Result:

project_name    sub_project_name        porosity
Bakken          Bakken Mylo QAQC        147
Bakken          Ghost fracture Study    252
Bossier         Doyle Boles K No.1      189
Eagleford       Kennedy Unit#1H         294
Eagleford       Wehmeyer Unit #1        0
Niobrara        Crow Valley 7-62-32-1M  0
Poland          Poland                  714
Woodford        Ridenour Phase 2        483
Woodford        Teague 1-14H            0

The correct result should be:

Bakken          Bakken Mylo QAQC        7
Bakken          Ghost fracture Study    12
Bossier         Doyle Boles K No.1      9
Eagleford       Kennedy Unit#1H         14
Eagleford       Wehmeyer Unit #1        0
Niobrara        Crow Valley 7-62-32-1M  0
Poland          Poland                  34
Woodford        Ridenour Phase 2        23
Woodford        Teague 1-14H            0

Answer 1

Try an explicit JOIN condition and explicit GROUP BY (although mysql allows to skip on the later).

SELECT p.project_name, sp.sub_project_name, COUNT(bl.bl_por) AS porosity
FROM   projects p
JOIN   sub_projects sp ON p.project_id = sp.projects_project_id
LEFT   JOIN bl ON sp.sub_project_id = bl.sub_project_id
GROUP  BY p.project_name, sp.sub_project_name;

---

Answer to additional question:

Again, try proper SQL syntax whit explicit JOIN and join-conditions:

SELECT p.project_name, sp.sub_project_name, COUNT(bl.bl_por) AS porosity
FROM   projects p
JOIN   sub_projects sp ON p.project_id = sp.projects_project_id
JOIN   wells w -- no JOIN condition? Results in cross join.
JOIN   cores c ON sp.sub_project_id = c.sub_projects_has_wells_sub_projects_sub_project_id
JOIN   samples s ON c.core_id = s.cores_core_id
JOIN   inputs i ON s.sample_id = i.samples_sample_id
LEFT   JOIN bl ON i.inputs_id = bl.inputs_inputs_id
GROUP  BY p.project_name, sp.sub_project_name

This way you notice immediately that the table wells is joined in without unconditionally. This results in a cross join: every row of the left side is extended with every row on the right side, which produces a lot of rows. Probably the source of your excessive counts.

Also be aware that count only counts non-null values. Any row where bl.bl_por IS NULL is not counted. If you actually want to count the number of rows, you can use count(bl.*).