Thank you, Zane, for sharing your thoughts ! Based on this input and positive responses from a few other folks (in irc or private mails), I'll start the official retirement process. On 1/17/24 11:52, Zane Bitter wrote:
On 20/12/23 22:10, Takashi Kajinami wrote:
Hello,
In heat project, we have had the heat-cfnclient library which contains the client utilities to consume AWS-compatible APIs in heat.
However this repository is not properly maintained and I doubt that anyone is using (or even can use) it with recent Heat.
- The repository still contains the utility for CloudWatch API which was already removed from heat
- The heat-cfn command relies on Keystone v2 instead of v3
- No release has been made since the repository was created
Based on the current situation I'd propose retiring the repository. I'd appreciate any feedback and if I hear no objections then I'll start the official process to retire the repository.
I think this makes sense.
IIRC the testing of heat-cfn-api is done through an independent client that doesn't rely on the actual heat-cfn client, right?
Yes, you are correct. We have a few tests of CFN API in heat-tempest-plugin but these tests use own implementation of client instead of heat-cfnclient.
I think it'd be worth removing heat-cfn-api altogether, but I guess there is still some need for it for signalling.
Yes. I'm also hoping to remove it, but the usage for signaling is the remaining concern as you said. I might look into the potential solution to retire it but I'll leave it for now as a separate topic.
(Historical note: the cfn-api *preceded* the native OpenStack API for Heat, but all innovation and development effort has gone into the native API since at least the beginning of 2013. The AWS CloudFormation client has hard-coded endpoints, and instead of forking it and making it configurable, we made heat-cfn-client as a boto wrapper instead.)
Ah, these makes the intention of the library very clear. Thanks for the explanation !
cheers, Zane.
Thank you, Takashi Kajinami