|
Bugzilla – Full Text Bug Listing |
| Summary: | [patch] Add support for non-unit cost metrics in routing | ||
|---|---|---|---|
| Product: | ns-3 | Reporter: | Tom Henderson <tomh> |
| Component: | routing | Assignee: | ns-bugs <ns-bugs> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | gjcarneiro |
| Priority: | P3 | ||
| Version: | pre-release | ||
| Hardware: | All | ||
| OS: | All | ||
| Attachments: | an exported hg changeset on top of changeset 1678 | ||
|
Description
Tom Henderson
2007-10-08 00:40:43 UTC
Created attachment 73 [details]
an exported hg changeset on top of changeset 1678
About the proposed interface: Ipv4Interface::SetMetric (uint16_t metric) Doesn't it make more sense to make it "int16_t metric", i.e. a signed integer? This way the default value could be set to zero, giving plenty of metric space either above or below the default value. Feels more natural, IMHO. Signed metric doesn't work for legacy routing protocols, including the global routing which is based on OSPF. I think the issue you are raising, though, is what are the semantics of this metric? Is it meant for all routing protocols, or just the global routing? Maybe it needs to be documented that this is a default metric but other routing protocols can override this in their own interface configuration files. applied as changeset 1776 |