html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,issue,performed_via_github_app
https://github.com/simonw/datasette/issues/670#issuecomment-849022714,https://api.github.com/repos/simonw/datasette/issues/670,849022714,MDEyOklzc3VlQ29tbWVudDg0OTAyMjcxNA==,9599,2021-05-26T18:33:47Z,2021-05-26T18:33:58Z,OWNER,Worth mentioning here: I've been doing a tun of research around running Datasette-like functionality against PostgreSQL in my https://github.com/simonw/django-sql-dashboard project - which will definitely inform the Datasette implementation.,"{""total_count"": 3, ""+1"": 3, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-797158641,https://api.github.com/repos/simonw/datasette/issues/670,797158641,MDEyOklzc3VlQ29tbWVudDc5NzE1ODY0MQ==,9599,2021-03-12T00:59:49Z,2021-03-12T00:59:49Z,OWNER,"> Challenge: what's the equivalent for PostgreSQL of opening a database in read only mode? Will I have to talk users through creating read only credentials?

It looks like the answer to this is yes - I'll need users to setup read-only credentials. Here's a TIL about that: https://til.simonwillison.net/postgresql/read-only-postgresql-user","{""total_count"": 1, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 1, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585962849,https://api.github.com/repos/simonw/datasette/issues/670,585962849,MDEyOklzc3VlQ29tbWVudDU4NTk2Mjg0OQ==,9599,2020-02-13T20:44:18Z,2020-02-13T20:46:53Z,OWNER,"Got the database page working! It lists tables, their columns and their row count.

Got the table page partially working! It can list rows.

It can't apply filters yet (because PostgreSQL `$1` parameters differ from SQLite `:blah` parameters) and faceting doesn't work because PostgreSQL requires that subqueries have an alias:
```
subquery in FROM must have an alias HINT: For example, FROM (SELECT ...) [AS] foo.
```
Still a pretty promising start though!","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585903240,https://api.github.com/repos/simonw/datasette/issues/670,585903240,MDEyOklzc3VlQ29tbWVudDU4NTkwMzI0MA==,9599,2020-02-13T18:28:42Z,2020-02-13T18:29:14Z,OWNER,Challenge: what's the equivalent for PostgreSQL of opening a database in read only mode? Will I have to talk users through creating read only credentials? Can I do this at runtime somehow? Can I detect if the connection has write permission and disable the arbitrary query feature?,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585885812,https://api.github.com/repos/simonw/datasette/issues/670,585885812,MDEyOklzc3VlQ29tbWVudDU4NTg4NTgxMg==,9599,2020-02-13T17:49:33Z,2020-02-13T17:49:33Z,OWNER,"```
In [12]: await conn.fetch(""""""SELECT a.attname, format_type(a.atttypid, a.atttypmod) AS data_type 
    ...: FROM   pg_index i 
    ...: JOIN   pg_attribute a ON a.attrelid = i.indrelid 
    ...:                      AND a.attnum = ANY(i.indkey) 
    ...: WHERE  i.indrelid = 'blog_blogmark'::regclass 
    ...: AND    i.indisprimary;"""""")                                                                                                                                   
Out[12]: [<Record attname='id' data_type='integer'>]
```","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585885242,https://api.github.com/repos/simonw/datasette/issues/670,585885242,MDEyOklzc3VlQ29tbWVudDU4NTg4NTI0Mg==,9599,2020-02-13T17:48:27Z,2020-02-13T17:48:27Z,OWNER,Finding out the primary keys for a table: https://wiki.postgresql.org/wiki/Retrieve_primary_key_columns,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585883109,https://api.github.com/repos/simonw/datasette/issues/670,585883109,MDEyOklzc3VlQ29tbWVudDU4NTg4MzEwOQ==,9599,2020-02-13T17:44:08Z,2020-02-13T17:44:08Z,OWNER,"Introspecting columns in PostgreSQL:
```
In [3]: sql = """"""SELECT * 
   ...:   FROM information_schema.columns 
   ...:  WHERE table_schema = 'public' 
   ...:    AND table_name   = 'blog_blogmark' 
   ...:      ;""""""                                                                                                                                                     
```
Each column looks like this:
```pythton
{'table_catalog': 'simonwillisonblog',
 'table_schema': 'public',
 'table_name': 'blog_blogmark',
 'column_name': 'id',
 'ordinal_position': 1,
 'column_default': ""nextval('blog_blogmark_id_seq'::regclass)"",
 'is_nullable': 'NO',
 'data_type': 'integer',
 'character_maximum_length': None,
 'character_octet_length': None,
 'numeric_precision': 32,
 'numeric_precision_radix': 2,
 'numeric_scale': 0,
 'datetime_precision': None,
 'interval_type': None,
 'interval_precision': None,
 'character_set_catalog': None,
 'character_set_schema': None,
 'character_set_name': None,
 'collation_catalog': None,
 'collation_schema': None,
 'collation_name': None,
 'domain_catalog': None,
 'domain_schema': None,
 'domain_name': None,
 'udt_catalog': 'simonwillisonblog',
 'udt_schema': 'pg_catalog',
 'udt_name': 'int4',
 'scope_catalog': None,
 'scope_schema': None,
 'scope_name': None,
 'maximum_cardinality': None,
 'dtd_identifier': '1',
 'is_self_referencing': 'NO',
 'is_identity': 'NO',
 'identity_generation': None,
 'identity_start': None,
 'identity_increment': None,
 'identity_maximum': None,
 'identity_minimum': None,
 'identity_cycle': 'NO',
 'is_generated': 'NEVER',
 'generation_expression': None,
 'is_updatable': 'YES'}
```","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585874699,https://api.github.com/repos/simonw/datasette/issues/670,585874699,MDEyOklzc3VlQ29tbWVudDU4NTg3NDY5OQ==,9599,2020-02-13T17:26:58Z,2020-02-13T17:27:13Z,OWNER,"First page to get working: the database view page, which shows a list of tables.

In the code this is entirely implemented with calls to the `Database` class, which is a good starting point - I can drop in a new version of that class that talks to PostgreSQL instead:

https://github.com/simonw/datasette/blob/0091dfe3e5a3db94af8881038d3f1b8312bb857d/datasette/views/database.py#L25-L66","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585873401,https://api.github.com/repos/simonw/datasette/issues/670,585873401,MDEyOklzc3VlQ29tbWVudDU4NTg3MzQwMQ==,9599,2020-02-13T17:24:38Z,2020-02-13T17:24:38Z,OWNER,"The biggest difference between the two will be around introspection. I searched the codebase for potential introspection queries, defined as `select ... from sqlite_master` and `pragma ...` statements.

- 7 results for sqlite_master in database.py
- 3 results in utils/__init__.py
- (Ignoring the 4 in inspect.py)
- 1 result for “pragma table_info” in app.py
- 2 in utils/__init__.py
- 2 results for “pragma foreign_key_list” in utils/__init__.py
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585872538,https://api.github.com/repos/simonw/datasette/issues/670,585872538,MDEyOklzc3VlQ29tbWVudDU4NTg3MjUzOA==,9599,2020-02-13T17:22:54Z,2020-02-13T17:22:54Z,OWNER,"A couple of things I'd like to support:
- The same datasette instance can have both PostgreSQL and SQLite databases attached to it, and both types will be listed on the homepage.
- The full test suite runs against both SQLite and PostgreSQL, with as few changes as possible (maybe a few `skipIf()` decorators for features not available on both). Probably easiest do do this with some kind of flag - e.g. `pytest --database=sqlite` v.s. `pytest --database=postgresql`

I can implement that with this in `conftest.py`:
```python
def pytest_addoption(parser):
    parser.addoption(""--database"", action=""store"", default=""sqlite"")
```
See https://stackoverflow.com/questions/40880259/how-to-pass-arguments-in-pytest-by-command-line","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,
https://github.com/simonw/datasette/issues/670#issuecomment-585870836,https://api.github.com/repos/simonw/datasette/issues/670,585870836,MDEyOklzc3VlQ29tbWVudDU4NTg3MDgzNg==,9599,2020-02-13T17:19:23Z,2020-02-13T17:19:37Z,OWNER,"I'm excited about https://github.com/MagicStack/asyncpg for this - it's a true async PostgreSQL library (my SQLite queries run in a threadpool right now) with extremely impressive performance benchmarks, from the team behind `uvloop`.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",564833696,