| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
RADIUS (RFC2865) is defined to be transported over UDP so the term
"radius packet" makes a lot of sense. RADIUS/TCP (RFC6613) and
RADIUS/TLS (RFC6614), a.k.a. RadSec, use stream transport protocols
though. The term "message" doesn't imply any kind of transport -- a
message can be sent using datagrams as well as in a stream.
This (large) commit changes 'package' to 'message' where it makes
sense. It does not touch the 'radius' subdirectory. It includes
preprocessor directives (#define) to make the public interface
compatible with previous releases of the library.
|
| |
|
|
| |
And distribute LICENSE and HACKING.
|
| |
|
|
| |
Fix from Luke Howard.
|
| | |
|
| |
|
|
|
|
|
| |
A new API for attributes will be added once we've decided how to deal
with RADIUS packets internally. For now, removing the half baked
wrapping seems more sensible than trying to free rs_attr objects and
their VALUE_PAIR's.
|
| |
|
|
| |
Also, move copyright notice out of COPYING and into every file.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
All aborts are removed, as well as all asserts which aren't
programming errors.
When an invalid packet is received, the connection is closed, as per
draft-ietf-radext-tcp-transport-08 (2.6.4).
Use new rs_debug() macro rather than fprintf() for debug printouts.
Coding style overhaul.
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
NOTE: Clean up of resources not yet sane. Expect resource leakages.
NOTE: Most failure cases are not handled properly. With the wind at
your back and the sun shining, it might work.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
