Accept postgresql://-schemed database URLs
#532
Conversation
Co-authored-by: Brandur Leach <[email protected]>
|
@brandur i had been waiting for the checks to turn green before requesting for another review, but for reasons i don’t know they’ve stayed amber. |
Ah yeah, GitHub requires us to approve CI runs from external contributors for security reasons (in case private vars or such leak into the build and could be extracted). |
|
Thanks! |
Prepare release v0.11.3, which includes largely #532, which is a legitimate bug in the CLI. This will be the first time I try for a release under workspaces (#528), so it includes a simultaneous release of the CLI and the rest of the project, although I'm expecting the build to fail due to lack of `replace` statements.
Prepare release v0.11.3, which includes largely #532, which is a legitimate bug in the CLI. This will be the first time I try for a release under workspaces (#528), so it includes a simultaneous release of the CLI and the rest of the project, although I'm expecting the build to fail due to lack of `replace` statements.
Prepare release v0.11.3, which includes largely #532, which is a legitimate bug in the CLI. This will be the first time I try for a release under workspaces (#528), so it includes a simultaneous release of the CLI and the rest of the project, although I'm expecting the build to fail due to lack of `replace` statements.
I ran into this as well a few weeks ago; I just changes postgresql:// to postgres://and called it a day 🙃 In-between doing a ton of other things and "just trying out this River thing to see if it works for us" (it does), I just never got around to reporting it. I would be surprised if I was the first person. I've often found that getting feedback on these types of small things can be very difficult. |
|
@arp242 Ah, fair enough. We used to have a rule of thumb at my last job that "for every one user that opened a support ticket, 10 users ran into the same problem and didn't". That's probably what happened here. |
Co-authored-by: Brandur Leach <[email protected]>
Prepare release v0.11.3, which includes largely riverqueue#532, which is a legitimate bug in the CLI. This will be the first time I try for a release under workspaces (riverqueue#528), so it includes a simultaneous release of the CLI and the rest of the project, although I'm expecting the build to fail due to lack of `replace` statements.
According to the official PostgresQL documentation (referenced 18/08/2024),
postgresql://is an acceptable database URL scheme. In fact that's what I use in my application. So I was surprised when River's migrations were failing. Turns out River only supportspostgres://-schemed URLs.Given that
postgresq://is accepted, and perhaps the preferred, I suggest that River accepts it to. Other changes are: fix grammar typo (unsupported instead of unsupport) and printing the URL the command was called with. I think it's good UX, especially if the user passed an environment variable.