paws-r / paws Goto Github PK

e.g. in paws.ec2::attach_volume.

# This example attaches a volume (``vol-1234567890abcdef0``) to an
# instance (``i-01474ef662b89480``) as ``/dev/sdf``.

Make sure generated packages list the right packages under Imports.

• Travis no longer works at all for this project.
• R CMD check notes and warnings don't trigger failures in Travis. (GitHub Actions replaced Travis)
• AppVeyor and Codecov always show up as "Queued" in the Checks tab for each PR.
• AppVeyor always builds make.paws against the master branch paws.common.
• Codecov stopped working after PR #86. It is a bug in script/codecov.R. See the Travis build log.

Travis for Linux, AppVeyor for Windows.

e.g. we should be able to set endpoints, etc. something like

config(
  Endpoint = "https://example.com"
)

IAM, CloudFront, etc. should sign requests from us-east-1 regardless which region the user chooses.

Implementation:

• Each service would be an exported object, e.g. paws::ec2.
• Each operation would be an element of that object, e.g. paws::ec2$run_instances.
• There would be a page in the documentation for each service, which links to its operations.
• Each operation's Roxygen name would be <service>_<operation>, e.g. ec2_run_instances, to avoid name collisions between services.
• Update the makefile so that make paws generates the single package.
• Update the readme and examples.

Benefits:

• Users need to download only one package.
• Only need to generate one package, and submit only one package to CRAN.
• Can run all integration tests with a single call to testthat. Would also potentially be able to get code coverage for the generated code.
• Don't need to fix issue #59, R package description warnings.

Costs:

• Documentation is harder to browse.
• Less idiomatic.

To-do:

• Finish fixing the HTML-to-markdown conversion, which still has bugs that lead to invalid R documentation.
• Add indexes listing all operations for each service. See tabular function reference.
• Update the examples & readme.
• Delete old packages and regenerate the new package.
• Add an argument to make_sdk() allowing one to specify which packages to build.
• Add tests for get_structure, read_api, make_sdk.
• Fix make_sdk bug s.t. if you put in a non-existent API, it doesn't fail, instead it builds the alphabetically first API.

In several packages, we get warnings about the description, e.g.

build paws.mediaconnect
Warning message:
'Description' must contain one or more complete sentences

e.g. Structure should enforce names, List enforce no names, etc.

From paws.ec2:

• checking DESCRIPTION meta-information (696ms) Dependence on R version '3.5.1' not with patchlevel 0
• checking Rd cross-references ... WARNING Missing link or links in documentation object 'modify_vpc_endpoint_service_permissions.Rd':
• checking for unstated dependencies in 'tests' ... WARNING 'library' or 'require' call not declared from: 'testthat'
• checking top-level files ... NOTE Non-standard file/directory found at top level: 'LICENSE.md'
• checking Rd files ... NOTE prepare_Rd: accept_reserved_instances_exchange_quote.Rd:36-38: Dropping empty section \examples

• Include documentation about nested elements, e.g. see boto3's run_instances.
• Show the full input shape in the usage section, including nested list elements. e.g. operation(Param = list(Sub = "string")).

HTML is not rendered inside code blocks, e.g. in iam_simulate_custom_policy, we currently get:

#' The ARN for an account uses the following syntax:
#' `arn:aws:iam::<i>AWS-account-ID</i>:root`. For example, to represent the

When making a request on an S3 bucket to an endpoint in the wrong region, you get a structured error response, of which Paws extracts and returns only the message. Full response below:

<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n
<Error>
  <Code>PermanentRedirect</Code>
  <Message>The bucket you are attempting to access must be addressed using the specified 
  endpoint. Please send all future requests to this endpoint.</Message>
  <Endpoint>s3.amazonaws.com</Endpoint>
  <Bucket>davidkretch</Bucket>
  <RequestId>12ABA16DB44973A8</RequestId>
  <HostId>i/luCtGDyDxKFG606J7t38Lc0r2aZUWnKCWgdPIZCNAzIYE6URwhaVh2j72ecaFNsCLUEWhV12M=</HostId>
</Error>

update_version.sh is supposed to update a package's version when there are changes, but instead it will always update, if only due to differences between Unix and Windows line breaks.

Check list for CRAN release:

• devtools::check()
• devtools::spell_check()
• devtools::check_rhub()
• devtools::check_win_devel()
• devtools::release()

In paws.elasticache operation copy_snapshot, "The S3 bucket %s does not exist." does not render correctly in the R help.

e.g. a scalar parameter should not receive a vector of length greater than one.

The populateutil functions should also enforce rules about their contents, e.g. Structure should require names, List should require no names, etc.

query_unmarshal_error does not work for S3 response errors, e.g.

(a <- put_bucket_tagging(
  Bucket = "Name-of-My-Bucket",
  Tagging = list(
    TagSet = list(
      list(
        Key = "Dog",
        Value = "Westie"
      ),
      list(
        Key = "Dog2",
        Value = "Beagle"
      )
    )
  )
))

The following packages receive default endpoints that do not exist.

paws.chime
paws.devicefarm
paws.globalaccelerator
paws.iot1clickdevicesservice
paws.kinesisanalyticsv2
paws.licensemanager
paws.mediastoredata
paws.migrationhub
paws.pinpointemail

Found the following (possibly) invalid file URI:
URI: NextToken
From: man/connect_get_current_metric_data.Rd

"NextToken" is not a file nor an URL.

Certain APIs require custom handlers. We should support adding them in from a file in make.paws/R/customizations.

a <- paws::ec2(config = list(region = "us-east-2"))
b <- paws::ec2(config = list(region = "us-east-1"))

# Returns the same list of key pairs.
a$describe_key_pairs()
b$describe_key_pairs()

Custom config code is on branch feature/custom_endpoint.

Code coverage links to files don't work.

See https://docs.codecov.io/docs/fixing-paths

Hi,
It seems that I can authenticate with normal AWS credentials but not with a temporary security credentials - is this correct and would it be possible to add support for this?

The following packages have integration test failures which also have XML error parse failures:

When running in the following code:

library(paws.batch)
(a <- describe_compute_environments(maxResults = 3))

I noticed that the parameters were not being used. The http request body is being created properly in the file handlers_jsonrpc.R.

However, the body is not being set at line 14 of that same file because request$client_info$target_prefix is null.

Hi all,
First off this is a great package and I really excited to started working with it. I am currently working on a wrapper of the python boto3 to create a R DBI interface into athena (https://github.com/DyfanJones/RAthena). I am really keen to do the same type of wrapper using the paws package as the driver package instead.

I just a quick questions first:

• Are you comfortable for the me start building the DBI wrapper package using paws as the driver? I am asking this as I don't want to interrupt/ stop future plans of the paws project.

Many thanks

e.g. the attributes we add to help unmarshal API responses.

paws.s3_service.R looks like this:

cfg <- client_config(
    service_name = "s3",
    endpoints = list("us-gov-west-1" = "s3.{region}.amazonaws.com", "us-west-1" = "s3.{region}.amazonaws.com", 
  )

Notice the list isn't completed

Issue

I'm trying to use paws package on an EC2 instance but it fails to get IAM credentials from the instance profile and i need to explicitly set the values for 'AWS_ACCESS_KEY_ID' and 'AWS_SECRET_ACCESS_KEY' in the environment. I get the following error when i try to get account summary on the instance or use any of the AWS services in the paws package with an IAM instance profile.

Below is the error message given for when i try to get account summary on an EC2 Instance

Error: Argument 'txt' must be a JSON string, URL or file.
Traceback:

1. temp$get_account_summary()
2. send_request(request)
3. sign(request)
4. run(request, sign)
5. handler$fn(request)
6. sign_sdk_request_with_curr_time(request)
7. sign_with_body(v4, request$http_request, request$body, name,
   . region, request$expire_time, request$expire_time > 0, signing_time)
8. get_credentials(signer$credentials)
9. provider()
10. get_iam_role()
11. jsonlite::fromJSON(iam_role_response$body)
12. stop("Argument 'txt' must be a JSON string, URL or file.")
13. base::stop(..., call. = FALSE)

iam_role_response is NULL which is why the error is given. If i do a simple GET request on the instance to the metadata url "http://169.254.169.254/....", i can successfully retrieve the security credentials which is where PAWS seems to be failing.

Steps to reproduce
Using R version 3.5.1, with latest PAWS package running on EC2 Instance

1. Paste the following code

Sys.setenv(
  AWS_REGION = "eu-west-2"
)

iaminfo <- paws::iam()
iaminfo$get_account_summary()

2. Run the code and wait a few seconds for the error message to appear.

Code generation takes a long time -- possibly multiple hours. The biggest part of code generation is converting HTML to Markdown or plaintext with Pandoc.