switch postgresql plugin driver - fixes poor handling of system column types.

[james-lawrence]

currently lib/pq doesn't handle name and oid types consistently making adding custom queries for other tables difficult. switching the driver resolves the issue.

I also just copied the helper code from lib/pq it was a dependency whose logic I didn't wish to change for this pr.

[sparrc]

@menardorama could you review?

[james-lawrence]

Just fyi, might be able to get away with just the pq driver, seems they fixed at least one of the datatypes that was a blocker.

[james-lawrence]

I just havent had the time to check

[sparrc]

@james-lawrence I would strongly prefer to keep the pq driver, does it only require updating the dependency?

[james-lawrence]

Depends. Not sure, assuming updating the dependency does resolve all the unsupported types then could remove some of the utility functions i had to add both for postgresql and the tests. But we'd want to keep the removal of the special cased columns.

[james-lawrence]

Willing to check next weekend.

[sparrc]

OK, I'm going to close the PR for the time being then while keeping the pq library is reviewed, feel free to re-open a new PR whenever you need

[james-lawrence]

@sparrc did some investigating this morning.

so libpq still doesn't handle datname properly. (its still returning a byte array instead of a string by default) but it does properly handle the column if you cast it in the query. (which it didn't in the past)
eg)
with libpq:

SELECT datid, datname::text, numbackends, xact_commit, xact_rollback, blks_read, blks_hit, tup_returned, tup_fetched, tup_inserted, tup_updated, tup_deleted, conflicts, temp_files, temp_bytes, deadlocks, blk_read_time, blk_write_time FROM pg_stat_database

with pgx:

SELECT * FROM pg_stat_database

pgx just handles datname and other special postgresql typed columns correctly which makes for a much nicer experience on the end user side.

I think its worth using pgx.

[sparrc]

@james-lawrence is this is a bug in the pq library? Have you looked into submitting a PR to fix this behavior in the pq library?

[james-lawrence]

I consider it a bug unsure if the maintainers of pq will. I'm not interested in putting in the time with pq to see if it will get fixed, in the past I've had trouble getting them to make changes that could impact end users codes. changing the return type for a postgresql column type would fall into that category.

[james-lawrence]

examples of pq unwilling to change returned types:
lib/pq#300
but feel free to report it to them and try to get them to change it =)