* upstream/master: (295 commits)
Update CHANGELOG.md
provider/aws: Allow DB Parameter group to change in RDS
return error if failed to set tags on Route 53 zone
core: [tests] fix order dependent test
Fix hashcode for ASG test
provider/aws: Fix issue with tainted ASG groups failing to re-create
Don't error when reading s3 bucket with no tags
Avoid panics when DBName is not set
Add floating IP association in aceptance tests
Use env var OS_POOL_NAME as default for pool attribute
providers/heroku: Add heroku-postgres to example
docs: resource addressing
providers/heroku: Document environment variables
providers/heroku: Add region to example
Bugfix on floating IP assignment
Update CHANGELOG.md
update CHANGELOG
website: note on docker
core: formalize resource addressing
core: fill out context tests for targeted ops
...
* master: (167 commits)
return error if failed to set tags on Route 53 zone
core: [tests] fix order dependent test
Fix hashcode for ASG test
provider/aws: Fix issue with tainted ASG groups failing to re-create
Don't error when reading s3 bucket with no tags
Avoid panics when DBName is not set
Add floating IP association in aceptance tests
Use env var OS_POOL_NAME as default for pool attribute
providers/heroku: Add heroku-postgres to example
docs: resource addressing
providers/heroku: Document environment variables
providers/heroku: Add region to example
Bugfix on floating IP assignment
Update CHANGELOG.md
update CHANGELOG
website: note on docker
core: formalize resource addressing
core: fill out context tests for targeted ops
core: docs for targeted operations
core: targeted operations
...
* upstream/master:
return error if failed to set tags on Route 53 zone
cleanups
provider/aws: Finish Tag support for Route 53 zone
provider/aws: Add tags to Route53 hosted zones
* master: (172 commits)
core: [tests] fix order dependent test
Fix hashcode for ASG test
provider/aws: Fix issue with tainted ASG groups failing to re-create
Don't error when reading s3 bucket with no tags
Avoid panics when DBName is not set
Add floating IP association in aceptance tests
Use env var OS_POOL_NAME as default for pool attribute
providers/heroku: Add heroku-postgres to example
docs: resource addressing
providers/heroku: Document environment variables
providers/heroku: Add region to example
Bugfix on floating IP assignment
Update CHANGELOG.md
update CHANGELOG
website: note on docker
core: formalize resource addressing
core: fill out context tests for targeted ops
core: docs for targeted operations
core: targeted operations
user_data support
...
* master: (172 commits)
core: [tests] fix order dependent test
Fix hashcode for ASG test
provider/aws: Fix issue with tainted ASG groups failing to re-create
Don't error when reading s3 bucket with no tags
Avoid panics when DBName is not set
Add floating IP association in aceptance tests
Use env var OS_POOL_NAME as default for pool attribute
providers/heroku: Add heroku-postgres to example
docs: resource addressing
providers/heroku: Document environment variables
providers/heroku: Add region to example
Bugfix on floating IP assignment
Update CHANGELOG.md
update CHANGELOG
website: note on docker
core: formalize resource addressing
core: fill out context tests for targeted ops
core: docs for targeted operations
core: targeted operations
user_data support
...
* d.Set has a pointer nil check we can lean on
* need to be a bit more conservative about nil checks on nested structs;
(this fixes the RDS acceptance tests)
/cc @fanhaf
s3.GetBucketTagging returns an error if there are no tags associated
with a bucket. Consequently, any configuration with a tagless s3 bucket
would fail with an error, "the TagSet does not exist".
Handle that error more appropriately, interpreting it as an empty set of
tags.
* f-aws-rds-tags:
fix index out of range error
fix formatting
upgrade VPC Ids and DB Subnet to be optionally computed
fix typo
provider/aws: Introduce IAM connection
* master:
provider/aws: Fix dependency violation when deleting Internet Gateways
command/remote-config: failing tests
update CHANGELOG
command/remote-config: do a pull with `terraform remote config`
command/remote-{pull,push}: colorize and show success output
command/remote-config: lowercase the type so that Atlas works, for example
command/remote-config: show flag parse errors
command/remote-config: remove weird error case that shows no error message
command: when setting up state, only write back if local is newer
* master: (66 commits)
provider/aws: Fix dependency violation when deleting Internet Gateways
command/remote-config: failing tests
update CHANGELOG
command/remote-config: do a pull with `terraform remote config`
command/remote-{pull,push}: colorize and show success output
command/remote-config: lowercase the type so that Atlas works, for example
command/remote-config: show flag parse errors
command/remote-config: remove weird error case that shows no error message
command: when setting up state, only write back if local is newer
minor code cleanups to get acceptance tests passing
update CHANGELOG
providers/digitalocean: add dot in GET response
providers/digitalocean: force fqdn in dns rr value
update CHANGELOG
small code cleanup
Add proper reading/updating of tags for S3
provider/aws: Add tags to S3
Documentation for ASG Tags added
Tags support added for AWS ASG
command/output: don't panic if no root module in state [GH-1263]
...
* master:
update CHANGELOG
providers/digitalocean: add dot in GET response
providers/digitalocean: force fqdn in dns rr value
update CHANGELOG
Add disk size to google_compute_instance disk blocks.
'project' should be set to the project's ID, not its name.
Don't error when enabling DNS hostnames in a VPC
Correct AWS VPC or route table read functions
Updates to GCE Instances and Instance Templates to allow for false values to be set for the auto_delete setting.
Update GCE Instance Template tests now that existing disk must exist prior to template creation.
Update Google API import to point to the new location.
add network field to the network_interface
I was working on building a validation to check the user-provided
"device_name" for "root_block_device" on AWS Instances, when I realized
that if I can check it, I might as well just derive it automatically!
So that's what we do here - when you customize the details of the root
block device, device name is just comes from the selected AMI.
The AWS API call ModifyVpcAttribute will allow only one attribute to be
modified at a time. Modifying both results in the error:
Fields for multiple attribute types specified: enableDnsHostnames, enableDnsSupport
Retructure the provider to honor this restriction.
Also, enable DNS support before attempting to enable DNS hostnames,
since the former is a prerequisite of the latter.
Additionally, fix what must have been a copy&paste error, setting
enable_dns_support to the value of enable_dns_hostnames.
If the state file contained a VPC or a route table which no longer
exists, Terraform would fail to create the correct plan, which is to
recreate them.
In the case of VPCs, this was due to incorrect error handling. The AWS
SDK returns a aws.APIError, not a *aws.APIError on error. When the VPC
no longer exists, upon attempting to refresh state Terraform would
simply exit with an error.
For route tables, the provider would recognize that the route table no
longer existed, but would not make the appropriate call to update the
state as such. Thus there'd be no crash, but also no plan to re-create
the route table.
Though not directly connected, trying to delete a subnet and security group in
parallel can cause a dependency violation from the subnet, claiming there are
dependencies.
This commit fixes that by allowing subnet deletion to tolerate failure with a
retry / refresh function.
Fixes#934
Instance block devices are now managed by three distinct sub-resources:
* `root_block_device` - introduced previously
* `ebs_block_device` - all additional ebs-backed volumes
* `ephemeral_block_device` - instance store / ephemeral devices
The AWS API support around BlockDeviceMapping is pretty confusing. It's
a single collection type that supports these three members each of which
has different fields and different behavior.
My biggest hiccup came from the fact that Instance Store volumes do not
show up in any response BlockDeviceMapping for any EC2 `Describe*` API
calls. They're only available from the instance meta-data service as
queried from inside the node.
This removes `block_device` altogether for a clean break from old
configs. New configs will need to sort their `block_device`
declarations into the three new types. The field has been marked
`Removed` to indicate this to users.
With the new block device format being introduced, we need to ensure
Terraform is able to properly read statefiles written in the old format.
So we use the new `helper/schema` facility of "state migrations" to
transform statefiles in the old format to something that the current
version of the schema can use.
Fixes#858
Fixes a bug in Route53 and wildcard entries. Refs #501.
Also fixes:
- an issue in the library where we don't fully wait for the results, because the
error code/condition changed with the migration to aws-sdk-go
- a limitation in the test, where we only consider the first record returned
* master:
provider/aws: Fix encoding bug with AWS Instance
minor style cleanups
Tags Schema
Added Tagging
Added vpc refactor in aws sdk go
Removed additional variable for print, added for debugging
Using hashicorp/aws-sdk-go
Changed things around as suggested by @catsby
Refactor with Acceptance Tests
VPC Refactor
First refactor
Added Connection to config
* master: (69 commits)
upgrade tests and remove ICMPTypeCode for now
helper/ssh: update import location
clean up
provider/aws: Convert AWS Network ACL to aws-sdk-go
Update website docs on AWS RDS encryption field
more test updates
provider/aws update Network ACL tests
code cleanup on subnet check
restore IOPS positioning
Code cleanup
Update CHANGELOG.md
Bugfix: Add tags on AWS IG creation, not just on update
fix nit-pick from go vet
remove duplicated function
provider/aws: Convert AWS Route Table Association to aws-sdk-go
Cleansup: Restore expandIPPerms, remove flattenIPPerms
clean up debug output to make go vet happy
providers/aws: Convert AWS VPC Peering to aws-sdk-go
provider/aws: Add env default for AWS_ACCOUNT_ID in VPC Peering connection
convert route table tests to aws-sdk-go
...
* master:
Code cleanup
Update CHANGELOG.md
fix nit-pick from go vet
remove duplicated function
provider/aws: Convert AWS Route Table Association to aws-sdk-go
Cleansup: Restore expandIPPerms, remove flattenIPPerms
clean up debug output to make go vet happy
providers/aws: Convert AWS VPC Peering to aws-sdk-go
provider/aws: Add env default for AWS_ACCOUNT_ID in VPC Peering connection
convert route table tests to aws-sdk-go
provider/aws: Convert AWS Route Table to aws-sdk-go
providers/aws: iops in root device skipped when output state
Give route table assoc it's own copy of this method for now
provider/aws: Convert Main Route Table assoc. to aws-sdk-go
aws/Route53 record creation timeout 10->30 mins
provider/aws: Convert AWS Security Group to aws-sdk-go
Fixing up the tests to make them pass correctly
Fixing a corner case while retrieving a template UUID
Adding tests and docs for the new VPN resources
Adding a few new resources
Removing `ForceNew` from `final_snapshot_identifier` - it's a parameter
that's _only_ passed during the DeleteDBInstance API call, so it's perfectly
valid to change the attribute for an existing DB Instance.
fixes#1138
This allows you to set lifecycle create_before_destroy = true
and fixes#532 as then we'll make a new launch config, change
the launch config on the ASG, and *then* delete the old launch
config.
Also tried adding tests which unfortunately don't seem to fail...
* master:
providers/aws: Convert Launch Configurations to awslabs/aws-sdk-go
update CHANGELOG
terraform: test post state update is called
command: StateHook for continous state updates
terraform: more state tests, fix a bug
state: deep copies are required
terraform: make DeepCopy public
state/remote: increment serial properly
state: only change serial if changed
terraform: call the EvalUpdateStateHook strategically
terraform: PostStateUpdate hook and EvalUpdateStateHook
- Remove check on password for AWS RDS Instance
- Update documentation on AWS RDS Instance regarding DB Security Groups
- Change error handling to check error code from AWS API [ci skip]
The `SourceDestCheck` attribute can only be changed via
`ModifyInstance`, so the AWS instance resource's `Create` function calls
out to `Update` before it returns to take care of applying
`source_dest_check` properly.
The `Update` function originally guarded against unnecessary API calls
with `GetOk`, which worked fine until #993 when we changed the `GetOk`
semantics to no longer distinguish between "configured and zero-value"
and "not configured".
I attempted in #1003 to fix this by switching to `HasChange` for the
guard, but this does not work in the `Create` case.
I played around with a few different ideas, none of which worked:
(a) Setting `Default: true` on `source_dest_check' has no effect
(b) Setting `Computed: true` on `source_dest_check' and adding a `d.Set`
call in the `Read` function (which will initially set the value to `true`
after instance creation). I really thought I could get this to work,
but it results in the following:
```go
d.Get('source_dest_check') // true
d.HasChange('source_dest_check') // false
d.GetChange('source_dest_check') // old: false, new: false
```
I couldn't figure out a way of coherently dealing with that result, so I
ended up throwing up my hands and giving up on the guard altogether.
We'll call `ModifyInstance` more than we have to, but this at least
yields expected behavior for both Creates and Updates.
Fixes#1020
library.
This commit updates the Route 53 Zone resource to use AWS Labs aws-sdk-go
library instead of mitchellh/goamz.
- hard code us-east-1 for Route53 region, since it's a global endpoint
- add some units test for CleanZoneID
Unfortunately, the acceptance tests here were improperly passing, and
allowing Subnet updates on ELBs is not as straightfoward as simply
removing `ForceNew`.
Subnets on ELBs need to be managed by two explicit API calls:
* `AttachLoadBalancerToSubnets` - http://bit.ly/elbattachsubnet
* `DetachLoadBalanceFromSubnets` - http://bit.ly/elbdetachsubnet
We'll need to circle back and use these APIs to explicitly add support.
This fixes the failure of `TestAccAWSELB_AddSubnet` by removing the
test.
This reverts commit 61e91017be, reversing
changes made to 49b3afe452.
Was relying on old behavior of GetOk and therefore never properly seeing
a change from true -> false.
This fixes the acceptance test failure of
`TestAccAWSInstance_sourceDestCheck`.
AWS provides a single `BlockDeviceMapping` to manage three different
kinds of block devices:
(a) The root volume
(b) Ephemeral storage
(c) Additional EBS volumes
Each of these types has slightly different semantics [1].
(a) The root volume is defined by the AMI; it can only be customized
with `volume_size`, `volume_type`, and `delete_on_termination`.
(b) Ephemeral storage is made available based on instance type [2]. It's
attached automatically if _no_ block device mappings are specified, and
must otherwise be defined with block device mapping entries that contain
only DeviceName set to a device like "/dev/sdX" and VirtualName set to
"ephemeralN".
(c) Additional EBS volumes are controlled by mappings that omit
`virtual_name` and can specify `volume_size`, `volume_type`,
`delete_on_termination`, `snapshot_id`, and `encryption`.
After deciding to ignore root block devices to fix#859, we had users
with configurations that were attempting to manage the root block device chime
in on #913.
Terraform does not have the primitives to be able to properly handle a
single collection of resources that is partially managed and partially
computed, so our strategy here is to break out logical sub-resources for
Terraform and hide the BlockDeviceMapping inside the provider
implementation.
Now (a) is supported by the `root_block_device` sub-resource, and (b)
and (c) are still both merged together under `block_device`, though I
have yet to see ephemeral block devices working properly.
Looking into possibly separating out `ephemeral_block_device` and
`ebs_block_device` sub-resources as well, which seem like the logical
next step. We'll wait until the next big release for this, though, since
it will break backcompat.
[1] http://bit.ly/ec2bdmap
[2] http://bit.ly/instancestorebytypeFixes#913
Refs #858
Right now we yield a perpetual diff on ASGs because we're not reading
termination policies back out in the provider.
This depends on https://github.com/mitchellh/goamz/pull/218 and fixes
it.
An `InstanceDiff` will include `ResourceAttrDiff` entries for the
"length" / `#` field of maps. This makes sense, since for something like
`terraform plan` it's useful to see when counts are changing.
The `DiffFieldReader` was not taking these entries into account when
reading maps out, and was therefore incorrectly returning maps that
included an extra `'#'` field, which was causing all sorts of havoc
for providers (extra tags on AWS instances, broken google compute
instance launch, possibly others).
* fixes#914 - extra tags on AWS instances
* fixes#883 - general core issue sprouted from #757
* removes the hack+TODO from #757
This resource allows an existing Route Table to be assigned as the
"main" Route Table of a VPC. This means that the Route Table will be
used for any subnets within the VPC without an explicit Route Table
assigned [1].
This is particularly useful in getting an Internet Gateway in place as
the default for a VPC, since the automatically created Main Route Table
does not have one [2].
Note that this resource is an abstraction over an association and does not
map directly to a CRUD-able object in AWS. In order to retain a coherent
"Delete" operation for this resource, we remember the ID of the AWS-created
Route Table and reset the VPC's main Route Table to it when this
resource is deleted.
refs #843, #748
[1] http://docs.aws.amazon.com/AmazonVPC/latest/UserGuide/VPC_Route_Tables.html#RouteTableDetails
[2] http://docs.aws.amazon.com/AmazonVPC/latest/UserGuide/VPC_Internet_Gateway.html#Add_IGW_Routing
If map_public_ip_on_launch was not specified, AWS picks a default of
"0", which is different than the "" in the state file, triggerinng an
update each time. Mark that parameter as Computed, avoiding the update.
This is necessary to support creating parameter groups with parameters
that require a reboot, since the RDS API will return an error when
attempting to set those parameters with ApplyMethod "immediate".
If a subnet exists in the state file and a refresh is performed, the
read function for subnets would return an error. Now it updates the
state to indicate that the subnet no longer exists, so Terraform can
plan to recreate it.
Several of the arguments were optional, and if omitted, they are
calculated. Mark them as such in the schema to avoid triggering an
update.
Go back to storing the password in the state file. Without doing so,
there's no way for Terraform to know the password has changed. It should
be hashed, but then interpolating the password yields a hash instead of
the password.
Make the `name` parameter optional. It's not required in any engine, and
in some (MS SQL Server) it's not allowed at all.
Drop the `skip_final_snapshot` argument. If `final_snapshot_identifier`
isn't specified, then don't make a final snapshot. As things were, it
was possible to create a resource with neither of these arguments
specified which would later fail when it was to be deleted since the RDS
API requires exactly one of the two.
Resolves issue #689.
- 5.6.17 is no longer a valid mysql engine version, bumping to 5.6.21
- updating security_group_names assertion to match new set structure
introduce in #663
When DeleteInternetGateway is successful it returns a nil error value.
However, for a nil error value, the RetryFunc returns an error yielding a
unnecessary second call to DeleteInternetGateway in the retry logic.
The logic works because DeleteInternetGateway eventually returns an ec2.Error
with error code InvalidInternetGatewayID.NotFound since the internet gateway
has been deleted in the previous call. The return value of nil breaks the
retry logic and the deletion is deemed successful.
Fix the unnecessary second call to DeleteInternetGateway by short circuiting
with a nil error value when deletion of the internet gateway is successful on
the first try.
Add an acceptance test for internet gateway deletion and remove unreachable
code while here.
If not suppling the `availability_zones`, they will be computed
(meaning an update/refresh will retrieve the info and update the values
to the state file).
So without the `Computed = true` the diff will always flag this as a
change, even when it’s not.
Some instance types have a block device by default. So when selecting
such an instance type, you will not set a config for the block device,
but the update/refresh func will notice one and update the state
nonetheless.
So in those cases the `block_device` becomes a `computed` field.
1. The schema contained a few fields that where not marked as
`computed`, while they were updated inside the resource.
2. While updating the `volume_size` it was doing so with a `string`,
but in the schema this field is set as `int`.
3. The set func for calculating the hashes for the `block` set items,
also used computed values to calculate the hash. As these values will
not be in the config, but only in the state, this will always show as a
diff. The solution is to only use the fields that aren’t computed in
order to get consistent hashes.
These where all issues before, but weren’t visible as such. All should
be good again now.
For now this only supports importing a key pair (by specifying a
public_key) property. In the future it'd be fairly trivial to support
key pair creation, with the private key returned as a computed property.
In real world usage you'd probably want to provide that public_key
property via a variable rather than hard-coding it into a terraform
config that'd end up in source control.
Fixes the following vet reports:
builtin/providers/aws/resource_aws_network_acl.go:191: wrong number of args for format in Errorf call: 2 needed but 3 args
builtin/providers/aws/resource_aws_network_acl.go:264: wrong number of args for format in Errorf call: 1 needed but 2 args
builtin/providers/aws/resource_aws_network_acl.go:268: wrong number of args for format in Errorf call: 1 needed but 2 args
builtin/providers/aws/resource_aws_network_acl.go:286: arg m[to_port].(int) for printf verb %s of wrong type: int
builtin/providers/aws/resource_aws_network_acl_test.go:277: arg r.NetworkAcls for printf verb %s of wrong type: []github.com/mitchellh/goamz/ec2.NetworkAcl
builtin/providers/aws/resource_aws_subnet_test.go:21: arg v.MapPublicIpOnLaunch for printf verb %s of wrong type: bool
There was an error in the goamz package the prevented updating the
availability zones correctly. So PR #181 should be merged before this
one can be merged…
By using a set for the availability zones, you can use things like
`availability_zones = ["${aws_instance.web.*.availability_zone}"]`
where is very likely multiple of the same zones will be added to the
set. If you use a list here, the list will say it’s changed (even if
you add the same zone) which will force a new resource.
The resource is build so it can attach and detach the Internet Gateway
from a VPC, but as the schema has `Required` and `ForceNew` both set
to `true` for the vpc_id field it will never use these capabilities.
This is a refactored solution for PR #616. Functionally this is still
the same change, but it’s implemented a lot cleaner with less code and
less changes to existing parts of TF.
Running the tests without these changes results in this error first:
```
--- FAIL: TestAccAWSNetworkAclsOnlyIngressRulesChange (24.92 seconds)
testing.go:121: Step 0 error: Check failed: Invalid number of ingress
entries found; count = %!s(int=3)
FAIL
exit status 1
FAIL github.com/hashicorp/terraform/builtin/providers/aws 24.974s
```
And after fixing that one you also get a few unexpected values due to
an expected order mismatch between the items in the set versus the
items in the config.
Those are also fixed, so the test is passing now.
Otherwise it is impossible to get simple configurations with the
"default" routing table, which is only the "local" route.
The following contents of main.tf expose the bug, and boots an instance
fine after this patch is applied:
variable aws_access_key {}
variable aws_secret_key {}
variable aws_ubuntu_ami {}
provider "aws" {
access_key = "${var.aws_access_key}"
secret_key = "${var.aws_secret_key}"
region = "us-east-1"
}
resource "aws_vpc" "default" {
cidr_block = "10.0.0.0/20"
}
resource "aws_route_table" "private" {
vpc_id = "${aws_vpc.default.id}"
# Note the lack of "route" sub-key here.
}
resource "aws_subnet" "private" {
vpc_id = "${aws_vpc.default.id}"
cidr_block = "10.0.1.0/24"
}
resource "aws_route_table_association" "private" {
subnet_id = "${aws_subnet.private.id}"
route_table_id = "${aws_route_table.private.id}"
}
# Demonstrate an instance can be booted fine in this fashion.
resource "aws_instance" "sample" {
ami = "${var.aws_ubuntu_ami}"
instance_type = "t2.micro"
subnet_id = "${aws_subnet.private.id}"
}
terraform.tfvars for completeness:
aws_access_key = "..."
aws_secret_key = "..."
# A public Trusty AMI
aws_ubuntu_ami = "ami-9aaa1cf2"
builtin/providers/aws/tags_test.go:56: unrecognized printf verb 'i'
builtin/providers/aws/tags_test.go:59: unrecognized printf verb 'i'
config/config_test.go:101: possible formatting directive in Fatal call
config/config_test.go:157: possible formatting directive in Fatal call
config/module/get_file_test.go:91: missing argument for Fatalf(%s): format reads arg 1, have only 0 args
helper/schema/schema.go:341: arg v.Type for printf verb %s of wrong type: schema.ValueType
helper/schema/schema.go:656: missing argument for Errorf(%s): format reads arg 2, have only 1 args
helper/schema/schema.go:912: arg schema.Type for printf verb %s of wrong type: schema.ValueType
terraform/context.go:178: arg v.Type() for printf verb %s of wrong type: github.com/hashicorp/terraform/config.VariableType
terraform/context.go:486: arg c.Operation for printf verb %s of wrong type: terraform.walkOperation
terraform/diff_test.go💯 arg actual for printf verb %s of wrong type: terraform.DiffChangeType
terraform/diff_test.go:235: arg actual for printf verb %s of wrong type: terraform.DiffChangeType
Prior to this, the diff only contained changed set elements. The issue
with this is that `getSet`, the internal function that reads a set from
the ResourceData, expects that each level (state, config, diff, etc.)
has the _full set_ information. This change was done to fix merging
issues.
Because of this, we need to make sure the full set is visible in the
diff.
cc/ @deoxxa see diff for how to add tests – quite straight forward.
Normally, I would like to add a check to ensure that the attribute
was updated remotely (by seeing what it's value with AWS is) but
I don't think we have that value returned.
When trying to output the `endpoint` attribute of an `aws_db_instance`,
the port was not being emitted properly. Converting the port number from
an integer to a string has resolved the issue.
cc/ @mitchellh
This one caught me a bit, and I'm not sure if there's another AttrType
that could be useful here – perhaps a AttrTypeHidden or something? I
don't know...this feels sloppy to have in the diff func.
/cc @pearkes - A few things here:
First, this fixes the issue you mentioned to be in chat. Basically: if
there are no security groups, don't put it into flatten, because flatten
will include a "key.#" = "0".
Next, I transformed your test into a "table-driven" test which are really
nice to write and maintain. Basically, as you find bugs or edge cases, you can
just add to the table and you get the test for free. I recommend these
whereever you have a pure input to output sort of data transformation
function.
Rui Zhang