to e2e, or not e2e, that is the question #22918
Conversation
versions The terraform version was hard-coded to 0.10.2, and the provider versions supported the older provider protocol version 4.
filepath.EvalSymlinks is our friend! The code already does this, the tests needed to be updated to do the same.
The DefaultFunc for insecure_skip_tls_verify was sending an empty string instead of a bool. Fixes to default to `false`.
ghost
added
the
sdkv1
| @@ -131,6 +131,9 @@ func (b *Backend) configure(ctx context.Context) error { | |||
| if data.Get("key_id").(string) == "" { | |||
| validationError = multierror.Append(validationError, errors.New("`Key ID` must be configured for the Triton provider")) | |||
| } | |||
| if data.Get("key_id").(string) == "" { | |||
| validationError = multierror.Append(validationError, errors.New("`Key ID` must be configured for the Triton provider")) | |||
There was a problem hiding this comment.
This seems the same as the block above it ... 🤔 is there a subtle difference I'm not seeing?
There was a problem hiding this comment.
Yes! The difference is "I need to update that with the correct attribute" 🤣 thank you so much
There was a problem hiding this comment.
On second thought, that shouldn't be in there at all - removed!
| @@ -19,7 +19,11 @@ func TestDirFromModule_registry(t *testing.T) { | |||
| } | |||
|
|
|||
| fixtureDir := filepath.Clean("testdata/empty") | |||
| dir, done := tempChdir(t, fixtureDir) | |||
| tmpDir, done := tempChdir(t, fixtureDir) | |||
| dir, err := filepath.EvalSymlinks(tmpDir) | |||
There was a problem hiding this comment.
Could you add a little comment above here explaining why this is necessary? It's surprising that a tempChdir result would need EvalSymlinks applied to it but I'm sure there was a reason! 😀
(Similarly for the other EvalSymlinks calls elsewhere here)
There was a problem hiding this comment.
Yes! Added a comment!
For PR posterity: The module installer runs filepath.EvalSymlinks on the destination directory, and the result of that is what's returned by the install hooks. If you are on, for eg, a mac whose TMPDIR env var points to a symlinked directory, the tests - which compare expected installation dir with what's returned by the install hooks - would fail.
I could also (easily) make the tests path by changing my machine's configured TMPDIR to a non-symlinked directory, but as I suspect this impacts most folks using mac osx and developing locally, I felt it was worth addressing in the test.
faster The acceptance tests for etcdv3, oss and manta were not validating required env variablea, chosing to assume that if one was running acceptance tests they had already configured the credentials. It was not always clear if this was a bug in the tests or the provider, so I opted to make the tests fail faster when required attributes were unset (or "").
mildwonkey
force-pushed
the
mildwonkey/to-e2e-or-not-e2e
branch
from
1c7a18d
to
68dfc30
Compare
This commit was generated from hashicorp/terraform#22918.
|
I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues. If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further. |
ghost
locked and limited conversation to collaborators
appilon
removed
the
sdkv1
While working on a new feature I found that certain e2e tests had not been updated for 0.12. This sent me down a bit of a rabbit trail getting the e2e tests to either pass or (in the case of backends) fail cleanly. (The first two commits represent tests I needed to get working for my feature while the latter two were fixed merely because the messiness of the failures annoyed me).
Some of these backend failures might represent bugs in the backend themselves (such as etcv3] but overall it seemed that the tests were simply making assumptions about env vars being set, then failing in weird ways when that was not the case.
I did find one clear bug in the
mantabackend: a bool attribute was defaulting to""and causing a panic, so I updated it tofalse(matching the documented default).