* Remove some unlikely to be used fields to help configurability.
* Combined "serverHost" and "serverPort" into "host"
* Remove "timeout" (just default to 30 seconds).
* Remove "maxIdleConn" will add it back if users feel the need
to control the number of cached connections.
* Remove "trustedEmailProvider" (just always trust).
* Remove "skipCertVerification" you can't make this connector
ingore TLS errors.
* Fix configs that don't search before bind (previously broken).
* Add more examples to Documentation
* Refactor LDAPPool Acquire() and Put() into a Do() function which
always does the flow correctly.
* Added more comments and renamed some functions.
* Moved methods on LDAPIdentityProvider to the LDAPConnector
Not all LDAP servers have entryDN available as an attribute. Reading up on
https://tools.ietf.org/html/rfc5020 tells me that entryDN is intended for
making the DN available for attribute value assertions. Thus it is not
mandatory for a LDAP server to make it available as an retrievable
attribute.
The DN is always a part of the entry returned in a search result, just use
it.
Fixes#314
Authentication is performed by binding to the configured LDAP server using
the user supplied credentials. Successfull bind equals authenticated user.
Optionally the connector can be configured to search before authentication.
The entryDN found will be used to bind to the LDAP server.
This feature must be enabled to get supplementary information from the
directory (ID, Name, Email). This feature can also be used to limit access
to the service.
Example use case: Allow your users to log in with e-mail address instead of
the identification string in your DNs (typically username).
To make re-use of HTTP form handling code from the Local connector possible:
- Implemented IdentityProvider interface
- Moved the re-used functions to login_local.go
Fixes#119
