NAME

Globus - Object-Oriented interface to Globus

DESCRIPTION

Globus is a tool that allows the sharing of scientific data between researchers and institutions. Globus enables you to transfer your data using just a web browser, or using their SSH interface at cli.globusonline.org.

This is a client library for the Globus CLI.

For detailed documentation of the API, see http://dev.globus.org/cli/reference.

CAVEATS

This code is a work in progress, focusing on my needs at the moment rather than covering all the capabilities of the Globus CLI. It is therefore very stubtastic.

This module also relies very much on SSH, and thus the rules of private and public keys. Therefore, using it as a shared tool would be ill-advised if not impossible.

SYNOPSIS

    my $g = Globus->new($username,$path_to_ssh_key) ;
    $g->endpoint_add_shared( 'institution#endpoint', $directory, $endpoint_name ) ;
    $g->acl_add( $endpoint . '/', 'djacoby@example.com' ) ;

METHODS

BASICS

new

    Creates a new Globus object. Takes two options: 
    the username and path to the SSH key you use to connect to Globus.

set_username

set_key_path

get_username

get_key_path

    These commands return and change the username and keypath you use to 
    connect to Globus.

TASK MANAGEMENT

cancel

details

events

modify

status

wait

We do not do much with task management, so these are currently stubs.

TASK CREATION

delete

rm

Currently stubs

scp

transfer

Both commands take a source, or from path (including endpoint), a destination, or to path (includint endpoint), and a boolean indicating whether you're copying recursively or not.

FILE MANAGEMENT

ls

Works?

rename

mkdir

Stubs

ENDPOINT MANAGEMENT

acl_add

acl_list

acl_remove

acl-* is the way that Globus refers to permissions

By the interface, Globus supports adding shares by email address, by Globus username or by Globus group name. This module sticks to using email address. acl_add() takes an endpoint, an email address you're sharing to, and a boolean indicating whether this share is read-only or read-write. acl_add() returns a share id.

acl_remove() uses that share id to identify which shares are to be removed.

acl_list() returns an array of hashes containing the information about each user with access to an endpoint, including the share ID and permissions.

endpoint_add_shared

endpoint_list

endpoint_search

endpoint_remove

endpoint_add_shared() handles the specific case of creating an endpoint from an existing endpoint, not the general case. It takes the endpoint where you're sharing from, the path you're sharing, and the endpoint you're creating. If you are user 'user' and creating the endpoint 'test', the command takes 'test', not 'user#test'.

endpoint_remove and endpoint_list, however, take a full endpoint name, like 'user#test'.

Current usage is endpoint_list for a list of all our shares, and endpoint_search for details of each individual share

list_my_endpoints

search_my_endpoints

list_my_endpoints() and search_my_endpoints() were added once I discovered the failings of existing list and search. These tools return a hashref of hashrefs holding the owner, host_endpoint, host_endpoint_name, credential_status, and most importantly, the id, legacy_name and display_name.

For older shares, legacy_name will be something like 'purduegcore#hr00001_firstshare' and display_name will be 'n/a', while for newer shares, legacy_name will be 'purduegcore#SAME_AS_ID' and display_name will be like older shares' legacy_name, 'purduegcore#hr99999_filled_the_space'. In both cases, the value you want to use to get details or to remove a share is the id, which is a UUID.

endpoint_activate

endpoint_add

endpoint_deactivate

endpoint_modify

endpoint_rename