When publish a techdocs to S3 all files has the content-type application/octet-stream. It does no break techdocs view inside backstage, but does not allow use a bucket as a static web site outside of backstage.

Expected Behavior

Send content-type as text/html for html files, text/css for css files, text/javascript for javascript files, application/xml for xml files and application/json for

Current behavior

Right now, the connection string to Azure can be passed as a string at initialization or read AZURE_STORAGE_CONNECTION_STRING from the environment.

The connection string property is not serialized with the storage object. The only way to get this to work is to have AZURE_STORAGE_CONNECTION_STRING available when the flow is retrieved from storage. For most agent types, t

What is the problem this feature would solve? Please describe.

In CI and/or scripting it is often useful for commands that create resources to block until they become ready. For example, both doctl compute droplet create and doctl kubernetes cluster create have --wait flags that block until the resource has been successfully created. Recently, the doctl apps create-deployment comman

🚀 Feature Request

The evaluate_loader method for Python API. Similar to .train and .predict_loader

Motivation

A number of modules, and scripts in the test suite start with

#!/usr/bin/env python

In general it is better to use

#!/usr/bin/env python

because that allows execution inside a virtualenv, or when python is installed elsewhere, e.g. /usr/loca/bin/python.

There may be cases in which we really do want to hard code the path - for instance testing detection/selection of `

https://www.usenix.org/conference/opml19/program

Prepare to read papers