You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Bug Report]: Publishing of resources with '--' in name fails (gracefully) #3592
- [digital-twins/digital-twins-instances/endpoints--service-bus] [latest]
VERBOSE: Invoke task with
VERBOSE: {
"TemplateFilePath": "/home/runner/work/ResourceModules/ResourceModules/modules/digital-twins/digital-twins-instances/endpoints--service-bus/main.bicep",
"BicepRegistryRgName": "artifacts-rg",
"BicepRegistryRgLocation": "West Europe",
"ModuleVersion": "latest",
"BicepRegistryName": "adpsxxazacrx001"
}
VERBOSE: Performing the operation "Publish" on target "Private bicep registry entry [bicep/modules/digital-twins.digital-twins-instances.endpoints--service-bus] version [latest] to registry [adpsxxazacrx001]".
The specified OCI artifact reference "br:adpsxxazacrx001.azurecr.io/bicep/modules/digital-twins.digital-twins-instances.endpoints--service-bus:latest" is not valid. The module path segment "digital-twins.digital-twins-instances.endpoints--service-bus" is not valid. Each module name path segment must be a lowercase alphanumeric string optionally separated by a ".", "_" , or "-".
VERBOSE: Publish complete
2 tasks:
Fix the publishing logic to fail the pipeline if the publishing fails.
Look at how to handle naming of resources when published to ACR (atleast). Other resources with the same issue:
changed the title [-][Bug Report]: Publishing of resources with '--' fails (gracefully)[/-][+][Bug Report]: Publishing of resources with '--' in name fails (gracefully)[/+]on Aug 20, 2023
Hey Erika Gressi (@eriqua), however we solve this - this should then also be consider for BRM if we ever get to publish child-modules.
There may be a fast workaround by updating the publishing in CARML to replace the -- with something that is allowed. However, I believe the 'correct' yet more intense solution would be to stop having different child-modules for the same sub-resource-type and instead have one with a corresponding User-Defined-Type interface for each collection of properties based on a discriminator. By now, this feature was added to the Bicep language, so we could implement it.
Alexander Sehr (@AlexanderSehr) sounds fair. Not sure what should be the best place for this issue though. l'd like to avoid implementing a convention in CARML and changing that later to align to BRM once we get to publishing children. I'm almost inclined to put this on hold until defining the convention in AVM. Meanwhile nothing prevents to test a possible implementation of course, but I wouldn't merge to main yet.
Activity
[-][Bug Report]: Publishing of resources with '--' fails (gracefully)[/-][+][Bug Report]: Publishing of resources with '--' in name fails (gracefully)[/+]ChrisGibson1982 commentedon Sep 25, 2023
Is there any resolution to this?
ChrisGibson1982 commentedon Sep 25, 2023
This is present in web\sites\hosting-environment as well
AlexanderSehr commentedon May 19, 2024
Hey Erika Gressi (@eriqua), however we solve this - this should then also be consider for BRM if we ever get to publish child-modules.
There may be a fast workaround by updating the publishing in CARML to replace the
--with something that is allowed. However, I believe the 'correct' yet more intense solution would be to stop having different child-modules for the same sub-resource-type and instead have one with a corresponding User-Defined-Type interface for each collection of properties based on a discriminator. By now, this feature was added to the Bicep language, so we could implement it.eriqua commentedon Jun 2, 2024
Alexander Sehr (@AlexanderSehr) sounds fair. Not sure what should be the best place for this issue though. l'd like to avoid implementing a convention in CARML and changing that later to align to BRM once we get to publishing children. I'm almost inclined to put this on hold until defining the convention in AVM. Meanwhile nothing prevents to test a possible implementation of course, but I wouldn't merge to main yet.