Closed shamrin closed 3 years ago
Hi @shamrin ,
- & 2. & 3.
The goal of triopg is to wrap all asyncpg features, while staying as close as possible to asyncpg api so that asyncpg's documentation stays relevant (the triopg documentation is basically it README, se we must keep things simple !). And as you discovered it's not that complicated to add wrappers around the add/remove methods ^^
I agree, a word about our ownconftest.py
in the README could help a lot
5.
In theory it shouldn't be very complicated to copy the function signature typing in the asyncpg api and add it to triopg. I guess we would be able to add this once asyncpg's PR has been merged
Does it make sense to have (some of) them in triopg? I can try to work on them.
As you demonstrated, README definitly needs improvements, you're PRs are welcome :+1:
Thank you for your response @touilleMan! I will try to improve the documentation.
(I still believe listen
helper would be very useful. It was not at all obvious how to implement it using asyncpg
API.)
implemented in #12 and #13 :smiley:
triopg
is a very nice library. It was a pleasure using it.However, there were a couple of places I got stuck when I started:
triopg
wraps most (all?) ofasyncpg
functionality. I even started making my own tinyasyncpg
wrapper before I've realized everything I need is available. Possible fixes: mention it in the README; write documentation.(see PR #12)
LISTEN
functionality. Asking on gitter helped though. Possible fix: add simplelisten
helper totriopg
:(see PR #13)
connect
does not sound very Trio-like. Possible fix: addopen_connection
oropen_conn
alias:It was not obvious how to write tests using real Postgres. I finally found that
triopg
own tests solve it pretty nicely. Possible fixes: document; providepytest
fixtures; send people totriopg
own test source code.Types. Should be easily solvable once MagicStack/asyncpg/pull/577 lands.
Does it make sense to have (some of) them in
triopg
? I can try to work on them.