fix: upgrade hasql-notifications to show error #3324
Conversation
steve-chavez
force-pushed
the
hasql-noti
branch
from
6d0757f
to
ba2d7b8
Compare
steve-chavez
force-pushed
the
hasql-noti
branch
from
ba2d7b8
to
2128f21
Compare
| # prev.callCabal2nixWithOptions "<name>" (super.fetchFromGitHub { | ||
| # lib.dontCheck (prev.callCabal2nixWithOptions "<name>" (super.fetchFromGitHub { | ||
| # owner = "<owner>"; | ||
| # repo = "<repo>"; | ||
| # rev = "<commit>"; | ||
| # sha256 = "<sha256>"; | ||
| # }) "--subpath=<subpath>" {}; | ||
| # }) "--subpath=." {}); |
There was a problem hiding this comment.
I don't think those changes make sense for the template here.
There was a problem hiding this comment.
Hm, why? I'm always lost when I see that <subpath> and lib.dontCheck is almost always the wanted behavior.
There was a problem hiding this comment.
While working on compiling with GHC 9.8 inside nix on a local branch, I have added plenty of those in the last couple of weeks. My experience so far:
lib.dontCheckwas used not very often - and even if it was, it should not be the default as in "copy & paste comes with it", but should be added once those tests actually fail.- Having
--subpath=.hides the fact that this is configurable and not some required default setting. I had quite a few cases where I needed a subpath other than.. On the other hand, if<subpath>is in there, you might need to think about it, yes, but the solution will be obvious, right?
There was a problem hiding this comment.
Ok cool. Addressed on #3326
| pkg = "hasql-notifications"; | ||
| ver = "0.2.1.0"; | ||
| sha256 = "sha256-MEIirDKR81KpiBOnWJbVInWevL6Kdb/XD1Qtd8e6KsQ="; |
There was a problem hiding this comment.
If we want to consider this as part of our bug fix, we should bump the lower bounds in postgrest.cabal to 0.2.1.0, too. This means we'd need to update stack.yaml with the extra-dep, too.
There was a problem hiding this comment.
Additionally the index-state in cabal.project.freeze needs to be updated, too.
There was a problem hiding this comment.
Hm, those are too easy to miss. Since we have cabal and stack running on CI, shouldn't those give an error?
There was a problem hiding this comment.
They won't be missed - if you take the first step before the second. Adjusting the package overlay for nix is the second step. If you need to bump dependencies, you should always go to postgrest.cabal first. Once you update the lower bound in there, everything else, including nix, will fail. But you need to update postgrest.cabal there is no way around that.
Based on diogob/hasql-notifications#21
The logs now show: