chore(deps): update hasql to 1.9.3.1 - #4829
Conversation
80433fb to
2e58d0f
Compare
| queryDbSettings preConfFunc prepared = | ||
| let transaction = if prepared then SQL.transaction else SQL.unpreparedTransaction in | ||
| transaction SQL.ReadCommitted SQL.Read $ SQL.statement dbSettingsNames $ SQL.Statement sql (arrayParam HE.text) decodeSettings prepared | ||
| SQL.transaction SQL.ReadCommitted SQL.Read $ SQL.statement dbSettingsNames $ SQL.Statement sql (arrayParam HE.text) decodeSettings prepared |
There was a problem hiding this comment.
Q: unpreparedTransaction went away? I remember we needed this because otherwise hasql prepared the BEGIN and COMMIT statements.
There was a problem hiding this comment.
Yeah, hasql-transaction v1.2 removed that: https://hackage.haskell.org/package/hasql-transaction-1.2.2/changelog
There was a problem hiding this comment.
I'm lost as where is this prepared bool now passed to prevent Hasql to prepare those TCL statements.
Although we have this test that should prove this behavior is maintained:
Lines 430 to 440 in bcc8998
There was a problem hiding this comment.
I don't quite get why I still pass prepared around here - but it seems to work. I did not look at the coverage report, though - there is probably a place somewhere, where this prepared thing is unused, and I should just remove it from here.
I had hoped for codecov to tell me, but somehow the job only fails, but doesn't post annotations? I'm not sure whether that's new, possibly since the latest updates, or when that started failing.
There was a problem hiding this comment.
I had hoped for codecov to tell me, but somehow the job only fails, but doesn't post annotations?
That's been going on for a while now, also miss it. I have to go to codecov UI to see them
There was a problem hiding this comment.
So passing prepared around is actually wrong here - and it has been before. This boolean is not really about whether to actually prepare the statement, but to tell hasql whether that statement is "preparable". Whether it is actually prepared or not is then determined by enabling prepared statements in the connection settings.
This means, I should just change all those instances of where prepared is still passed around to constant booleans (aka True).
There was a problem hiding this comment.
I had hoped for codecov to tell me, but somehow the job only fails, but doesn't post annotations?
That's been going on for a while now, also miss it. I have to go to codecov UI to see them
I've tried e2a677e, let's see what that does.
But in general, the future of annotations for codecov doesn't look bright - they are deprecated: https://docs.codecov.com/docs/github-checks - the same site also mentions some codecov browser extension one can use: https://docs.codecov.com/docs/the-codecov-browser-extension
2e58d0f to
a0eb26b
Compare
|
This is now only the relevant commit, the other PR has been merged. Will let that one sit for an actual review, since there are code changes (all the other PRs so far were only CI / nix stuff, so CI gave the feedback I needed). |
26c1226 to
605625b
Compare
605625b to
1a652ec
Compare
Michael and I arrived at mostly the same things independently. Took some of his, some of mine. Co-authored-by: Michal Kleczek <michal@kleczek.org>
1a652ec to
6b73b00
Compare
|
No other comments came up and there is already more work lining up behind this (#4222), so merging. |
3 participants
I originally had most of these changes in #4193 and some more on a local GHC 9.12 branch. @mkleczek came to many of the same changes in #4797. I took some of mine, some of his.
(first commit is still from #4193, only the second commit is relevant here)