Lix and CppNix disagree on how ?ref=
in flake inputs should be represented in lockfiles #520
Labels
No labels
Area/build-packaging
Area/cli
Area/evaluator
Area/fetching
Area/flakes
Area/language
Area/profiles
Area/protocol
Area/releng
Area/remote-builds
Area/repl
Area/store
bug
crash 💥
Cross Compilation
devx
docs
Downstream Dependents
E/easy
E/hard
E/help wanted
E/reproducible
E/requires rearchitecture
imported
Needs Langver
OS/Linux
OS/macOS
performance
regression
release-blocker
RFD
stability
Status
blocked
Status
invalid
Status
postponed
Status
wontfix
testing
testing/flakey
ux
No milestone
No project
No assignees
3 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference: lix-project/lix#520
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Describe the bug
If a
github
flake input URL contains a?ref
argument, CppNix will represent it as aref
in theflake.lock
, but Lix represents it as arev
in theflake.lock
. This leads to incompatibility when CppNix and Lix users attempt to collaborate on a project.Workaround
Replacing
ref
withrev
in the input URL seems to work:Then, Lix and CppNix will both represent it as a
rev
in theflake.lock
.Steps To Reproduce
Notes
For a
git+ssh://
URL, both CppNix and Lix only accept a?rev=
argument, and?ref=
fails with a rather odd error:This is most likely a CppNix regression; can you check if it works on 2.18? I have pretty high confidence we didn't break it since 2.18, but it's possible we did. i am also pattern matching this change as possibly related to fixing it: https://gerrit.lix.systems/c/lix/+/1353owwww i really can't read today sorry
As for the ssh URLs you're using the wrong URL format (git will also fail at this ssh url itself!), you need to replace : with / because : is for the port. The scp style ssh URLs are kind of a weird legacy thing in git itself that annoyingly don't support port numbers among other issues.
maybe @ma27 knows about/wants to see this bug
Most likely a similar issue as #472, I'm pretty sure it's the same root cause.
I'll see if I find a reasonable way to revert the behavior here.
But bottom line is that we shouldn't write lock files unless we actually need to.
Self assigning, will try to get something done within the next ~2 weeks (together with #472).