Added optional module names as an experimental feature

[majastrz]

Overview

Currently, module names are a required property on module declarations. Modules are compiled to nested deployments in the JSON and the name expression is transpiled as-is. With deeply-nested module hierarchies it is very easy to create module names that conflict in grandchild deployments, which causes difficult to diagnose failures at deploy time. This partially fixes #3897.

I'm introducing an experimental feature that once enabled, makes the name property in module declarations optional. There is no change in behavior if the feature is disabled.

With the feature enabled and name omitted in a module, we generate the following name expression in the JSON:

• In a module loop -> [format('<symbolic name prefix>-{0}-{1}', copyIndex(), uniqueString('<symbolic name>', deployment().name))]
• In a single module -> [format('<symbolic name prefix>-{0}', uniqueString('<symbolic name>', deployment().name))]

above represents the full symbolic name of the enclosing module declaration. is a substring of that starts at index 0 and takes as many characters as possible to fit within a deployment name.

Considerations

• The uniqueString() function always returns 13 characters
• We assume that the length of copyIndex() serialized to a string is always 3 characters since the max value is 800. While this approach leaves up to 2 characters unused in module loops with 1 or 2-digit iteration counts, it makes the symbolic name prefix consistent, improving readability in the portal.
• The full symbolic name and current deployment name are hashed into the module name to reduce the possibility of clashes. Chaining of deployment name hashes also avoids clashes with grandchild deployments using duplicated symbolic names.

Bicep authors should not take a dependency on the algorithm used to calculate the name. We reserve the right to change it at any time in the future without notice.

We will add a linter for module names in the future, but it is out of scope of this PR.

Portal experience

Module loop

main.bicep

param count int

module veryLongSymbolicNameThatExceedsTheMaximumPrefixLengthLimitSoICanTestMyAlgorithm 'mod.bicep' = [for i in range(0, count): {
  params: {
    templateSpecName: 'ts-${i}'
  }
}]

mod.bicep

param templateSpecName string

resource ts 'Microsoft.Resources/templateSpecs@2022-02-01' = {
  name: templateSpecName
  location: resourceGroup().location
}

Azure Portal (count = 500)

Module nesting

main.bicep

module veryLongSymbolicNameThatExceedsTheMaximumPrefixLengthLimitSoICanTestMyAlgorithm 'mod.bicep' = {
  params: {
    name: 'one'
  }
}

module veryLongSymbolicNameThatExceedsTheMaximumPrefixLengthLimitSoICanTestMyAlgorithm2 'mod.bicep' = {
  params: {
    name: 'two'
  }
}

mod.bicep

param name string

module veryLongSymbolicNameThatExceedsTheMaximumPrefixLengthLimitSoICanTestMyAlgorithm 'mod2.bicep' = {
  params: {
    name: name
  }
}

mod2.bicep

param name string

resource ts 'Microsoft.Resources/templateSpecs@2022-02-01' = {
  name: name
  location: resourceGroup().location
}

Azure Portal (count = 500)

Microsoft Reviewers: Open in CodeFlow

[github-actions]

Test this change out locally with the following install scripts (Action run 7069477119)

VSCode

• Mac/Linux

  bash <(curl -Ls https://aka.ms/bicep/nightly-vsix.sh) --run-id 7069477119

• Windows

  iex "& { $(irm https://aka.ms/bicep/nightly-vsix.ps1) } -RunId 7069477119"

Azure CLI

• Mac/Linux

  bash <(curl -Ls https://aka.ms/bicep/nightly-cli.sh) --run-id 7069477119

• Windows

  iex "& { $(irm https://aka.ms/bicep/nightly-cli.ps1) } -RunId 7069477119"

[majastrz]

module_symbolic_name_with_a_super_long_name_that_has_seventy_seven_characters

Guess how long this string is :D

[shenglol]

42

[majastrz]

:D

[github-actions]

Test Results

     143 files  +     11       143 suites  +11   5h 49m 2s ⏱️ + 1h 15m 4s
10 870 tests +     10  10 868 ✔️ +       8  0 💤 ±0  2 ❌ +2 
54 533 runs  +2 170  54 531 ✔️ +2 168  0 💤 ±0  2 ❌ +2 

For more details on these failures, see this check.

Results for commit 776b94c. ± Comparison against base commit edf1324.

♻️ This comment has been updated with latest results.

[FredericGobert]

@majastrz @shenglol Any idea how long this feature is going to be labeled as expiremental? It's getting close to a year since this feature was released, or are there situations where the feature still produces unwanted results?