Recover lost typings #2
Comments
|
We should probably do this soon to avoid heavy churn in all consumers later on. |
|
We've stayed in a middle ground here where most stuff doesn't add strongly typed wrappers, but a few things do. I actually wonder whether we want to just go back to not trying to project strong types onto the raw projections. That will simplify a lot of these cases, and give us (and users) a lot more flexibility. It's not quite as pretty, but in practice as we've built larger things (my work on containers, @mmdriley 's work on porting our CFN template, and @joeduffy + @mmdriley work on LM code), this hasn't felt like a cost (and indeed has helped ensure we don't hit too many roadblocks). Does it make sense to go back to a simple(r) project of Terraform here, and lave strong typing for folks who want to build components and layers on top? See also pulumi/pulumi#83. |
|
Let's decide what to do here as part of #130. |
|
We have a plan for how to approach this as part of https://github.com/pulumi/design/blob/master/meetings/2018/2018-02-20.md#pulumiaws. Given that we can add this strong typing without breaking changes - we will take these on a more case-by-case basis. |
|
At this point, this is tracking finding some way to proactively identify and map strongly-typed proeprty references for AWS and TF providers generally. Frankly, we've heard very little user feedback on this, and through many rounds of discussion have come up with no great ideas for how to get to "great" coverage on strongly typed references - so our investments have been very incremental, leading to unpredictable ability to pass strongly typed references. At this point, we really need to proposal for how to get to (and stay at) 90%+ coverage on strongly typed references or I would suggest we should not be trying to drive this further - as being at 50% coverage is likely worse than being at ~0% in terms of API consistency. |
|
I'm going to close this out - we've done a lot of work in and around this - and we'll continue to do more. But we have no near-term plan to do anything radically different in a structured way. And we broadly are adding these as incremental additions that preserve compatibility. |
# This is the 1st commit message: Fix import resources with provider default tags We have special logic around applying default provider tags to resources. This logic only applied to the `Check` call which means it was not applied when you were importing resources. This PR extends that logic to also run during the `Read` call. fix #4030, fix 4080 # This is the commit message #2: skip test # This is the commit message #3: fixing test # This is the commit message #4: Adding more tests # This is the commit message #5: Upgrade pulumi-terraform-bridge to v3.86.0 (#4160) This PR was generated via `$ upgrade-provider pulumi/pulumi-aws --kind=bridge --pr-reviewers=guineveresaenger`. Fixes #4091 Fixes #4137 --- - Upgrading pulumi-terraform-bridge from v3.85.0 to v3.86.0. - Upgrading pulumi-terraform-bridge/pf from v0.38.0 to v0.39.0. # This is the commit message #6: chore: run upstream provider-lint (#4120) This adds a step for running the upstream `provider-lint` make target. As part of this I had to fix some of the patches which violated some lint rules. **0009-Add-ECR-credentials_data_source.patch** - `ForceNew` does not apply to data sources **0032-DisableTagSchemaCheck-for-PF-provider.patch** - Schema have to have a `Type` - Also needed to add a ignore for `S013` which forces `Computed`, `Optional` or `Required` to be set. Looks like it can't recognize the `tagsComputed` var **0034-Fail-fast-when-PF-resources-are-dropped.patch** - Added a lint ignore for a rule which doesn't allow panics **0050-Normalize-retentionDays-in-aws_controltower_landing_.patch** - This test doesn't actually need a region or partition so replacing with a placeholder closes #4110 # This is the commit message #7: fix: CVE-2024-24791 (#4175) Fixes #4163 Upgrades minimally required Go versions to those unaffected by CVE-2024-24791.
After switching to the Terraform provider, we lost a fair number of strong typing niceties. Once we pop up for air, we should attempt to recover as many of them as possible. See the commit pulumi/pulumi@5eb8edc for a good reference of what it used to look like. Essentially any enums in that file are missing, and I'm sure many of the "capability-like" views have been lost in favor of weak IDs.
The text was updated successfully, but these errors were encountered: