Problems when installing package devtools and AzureML

[ActuarialPaul]

Dear friends,

In R 3.3.2 I ran the following code:

if(!require("devtools")) install.packages("devtools")
devtools::install_github("RevolutionAnalytics/azureml")

And this generated the following error:

devtools::install_github("RevolutionAnalytics/azureml")
Error in curl::curl_fetch_disk(url, x$path, handle = handle) :
Couldn't connect to server

When trying the same in R 3.1.0 and R 3.1.3, the following error is generated:

Installing AzureML
Error in _digest(c(list(repos, type), lapply(_additional, function(x) eval(x[[2L]], :
object 'digest_impl' not found

I really don´t know what is going on.

Can somebody give me some help with this?

Best regards.

[andrie]

I don't know the answer to your question, but is there a reason you're not installing from CRAN?

[akzaidi]

The initial error seems to be from curl. I'd update curl and httr and try again. But you should just try and use the CRAN version, since the latest version is on CRAN anyway.

[singlis]

That sounds like an issue with either curl installation or the way the proxy is setup for R.
Can you also try the devtools github sync on another project to see if it repros?

[ActuarialPaul]

Dear Scott, How do I configure the proxy for R? I talked to the network security people from our IT department and I was given a static IP address and was also given open connection to internet (so I can go out without going through any proxies), shouldn´t this be enough? Also, Andrie and Ali suggested to use the CRAN Version, but I don´t quite understand. I download all of my R software from the following link: https://cran.r-project.org/bin/windows/base/old/3.1.3/ https://cran.r-project.org/bin/windows/base/ I always use cran to download my R software so not sure what they mean by download from CRAN directly? Any additional guidance will be greatly appreciated, Thank you guys (Scott, Andrie and Ali), Best regards, Paul 2017-03-07 13:00 GMT-05:00 Scott Inglis <notifications@github.com>:

…

That sounds like an issue with either curl installation or the way the proxy is setup for R. Can you also try the devtools github sync on another project to see if it repros? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#123 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AXyNe_W_quUUyYL41RRkeqG1h4Frymdmks5rjZtUgaJpZM4MVkWM> .

[ActuarialPaul]

Hello guys, First of all, thank you for taking some time to guide me. I think there is something going on with RStudio and the function PublishWebService, I was looking at the following page: https://www.theta.co.nz/news-blogs/tech-blog/creating-smart-reports-and-applications-with-machine-learning-and-r I tried to recreate the simple example, just to check that a web service could be generated through RStudio and be consumed in Azure ML Studio, So here is the code I used: install.packages("RCurl") update.packages("RCurl") library("RCurl") install.packages("httr") update.packages("httr") library("httr") install.packages("RJSONIO") library("RJSONIO") install.packages("uuid") library("uuid") install.packages("jsonlite") library("jsonlite") install.packages("codetools") library("codetools") install.packages("base64enc") library("base64enc") install.packages("data.table") library("data.table") install.packages("df2json") library("df2jason") install.packages("devtools") library("devtools") install.packages("AzureML") library(AzureML) wsID2<-'6357fcb0c4454a2e97293589d6796f4d' authToken2<-'kNB6DOcS3RUPrn0b15s+CCsQIaSoWyxT0NVn4OcfoGdqakzc344L1NA7HvNJS1rTUNI7EmyqHJJF89sPXQCcFA==' add<-function(x,y) {return(x+y)} newService<-PublishWebService('add', 'add_data', list('x'='int', 'y'='int'), list('z'='int'), wsID2, authToken2) I was able to successfully install and load all the required packages, however the problem starts when I try to use the PublishWebService function (which supossedly is an RStudio function): add<-function(x,y) {return(x+y)}

newService<-publishwebService('add', 'add_data', list('x'='int',

'y'='int'), list('z'='int'), wsID2, authToken2) Error: could not find function "publishwebService"

newService<-PublishWebService('add', 'add_data', list('x'='int',

'y'='int'), list('z'='int'), wsID2, authToken2) Error: could not find function "PublishWebService"

newService<-PublishWebService('add', 'add_data', list('x'='int',

'y'='int'), list('z'='int'), wsID2, authToken2) Error: could not find function "PublishWebService"

Any guidance will be greatly appreciated, Cheers, Paul 2017-03-07 12:41 GMT-05:00 Andrie de Vries <notifications@github.com>:

…

I don't know the answer to your question, but is there a reason you're not installing from CRAN? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#123 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AXyNe5EgfUN1O1jPOoAnjMFdP2qISpAUks5rjZbOgaJpZM4MVkWM> .

[ActuarialPaul]

Hello again guys, Another question is, where do I find the ~/azureml/setings.json file? Cheers, Paul 2017-03-07 12:41 GMT-05:00 Andrie de Vries <notifications@github.com>:

…

I don't know the answer to your question, but is there a reason you're not installing from CRAN? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#123 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AXyNe5EgfUN1O1jPOoAnjMFdP2qISpAUks5rjZbOgaJpZM4MVkWM> .

[singlis]

Hello Paul,

I will try to repro your script. As a side note, I noticed the auth token in the code example above. I highly recommend changing your authentication token at the azure portal since this is a public forum.

As for the proxy setting, you can check to see if the following is set:

Sys.getenv("http_proxy")
Sys.getenv("https_proxy")

If any of those are set, then you are using a proxy server. They can be changed with Sys.setenv:
Sys.setenv(http_proxy="url or blank")

As for the settings file, you will need to create this. If you look in the documentation, "Using a settings.json" it gives an example for setting up the workspace. However, I pulled this from the example_datasets.R:

  # Use the default config file ~/azureml/settings.json with format:
  #   {"workspace":{
  #     "id":"test_id",
  #     "authorization_token": "test_token",
  #     "api_endpoint":"api_endpoint",
  #     "management_endpoint":"management_endpoint"
  #    }}
  # or, optionally set the `id` and `auth` parameters in the workspace
  # function.

  ws <- workspace()

Calling workspace() should load from that default location.

[ActuarialPaul]

Thank you so much, Where exactly do I change my auth token? Cheers, Paul 2017-03-07 15:29 GMT-05:00 Scott Inglis <notifications@github.com>:

…

Hello Paul, I will try to repro your script. As a side note, I noticed the auth token in the code example above. I highly recommend changing your authentication token at the azure portal since this is a public forum. As for the proxy setting, you can check to see if the following is set: Sys.getenv("http_proxy") Sys.getenv("https_proxy") If any of those are set, then you are using a proxy server. They can be changed with Sys.setenv: Sys.setenv(http_proxy="url or blank") As for the settings file, you will need to create this. If you look in the documentation, "Using a settings.json" it gives an example for setting up the workspace. However, I pulled this from the example_datasets.R: Use the default config file ~/azureml/settings.json with format: {"workspace":{ "id":"test_id", "authorization_token": "test_token", "api_endpoint":"api_endpoint", "management_endpoint":"management_endpoint" }} or, optionally set the id and auth parameters in the workspace function. ws <- workspace() Calling workspace() should load from that default location. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#123 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AXyNe2ya97eY-WZb8kwwp7UY9mxGOigLks5rjb4ogaJpZM4MVkWM> .

[singlis]

1. Login to azure.microsoft.com and find you Azure ML resource. When you select this, it will expand to three options.
2. Select "Launch Azure Machine Learning Studio". This will navigate to a new site.
3. Click "Sign In".
4. On the left side, click "Settings" then the "Authorization Token" tab.
5. Click "Regenerate" for each of the tokens.

These tokens you will now need to use in your script (really the first one should be fine, both are not needed for your script).

[singlis]

Hi Paul,

Very sorry for the delay in response. I had a chance to run through the scenario on a clean windows machine.

So I see you are trying out the example from the Getting Started - which is great. publishWebService is the correct call to be using (not PublishWebService).

Couple things that I noticed:

1. I would not use the blank ws, use the one that you created with your id and token. The example in the documentation creates a default one (ws <- workspace()), which assumes you have this configuration saved to disk.
2. I was missing zip on my machine. I had to install RTools (separate install from install.packages) that I downloaded from here:
   https://cran.r-project.org/bin/windows/Rtools

Make sure that you add RTools to your path through the install.

Here is what I ran, note (enter workspace id and enter workspace auth) need to be replaced.

> Sys.which('zip')
                       zip 
"c:\\Rtools\\bin\\zip.exe" 
> library(AzureML)
> ws <- workspace('<enter workspace id>', '<enter workspace auth token>')
> add <- function(x, y) {
+     x + y
+ }
> api <- publishWebService(
+     ws,
+     fun = add, 
+     name = "AzureML-vignette-silly",
+     inputSchema = list(
+         x = "numeric", 
+         y = "numeric"
+     ), 
+     outputSchema = list(
+         ans = "numeric"
+     )
+ )
> class(api)
[1] "Endpoint"   "data.frame"
> names(api)
 [1] "Name"                  "Description"           "CreationTime"          "WorkspaceId"           "WebServiceId"         
 [6] "HelpLocation"          "PrimaryKey"            "SecondaryKey"          "ApiLocation"           "GlobalParameters"     
[11] "MaxConcurrentCalls"    "DiagnosticsTraceLevel" "ThrottleLevel"  

Let me know how this works out. If I don't hear from you in the next couple of days, I'll close this issue.