[olsr-dev] new ipc

[Bernd Petrovitsch]

On Tue, 2006-03-07 at 16:45 +0100, Lorenz Schori wrote:
[...]
> ok i see. you are tending towards some sort of http like  
> implementation. perhaps using plain http would be not so stupid at  
> all because of it's availability.
> 
> "GET /capability" -> return protocol (core + plugin) names and  
> available fields. eg:
> route: state,destination,gateway,metric,etx,interface,type
> link: state,localip,remoteip,hysteresis,linkquality,lost,total,nlq,etx
> ...
> 
> perhaps the client knows exactly what it wants:
> "GET /?route=destination+gateway+etx&link=localip,remoteip,etx
                          ^       ^                ^        ^
I don't care which letter it is but it should be consistent.

> route: xx.xxx.xx.xxx xx.xxx.xx.xx x.xx
> route: yy.yyy.yy.yyy yy.yyy.yy.yy y.yy
> link: zz.zzz.zz.z zz.zzz.zz.zz zz.zz
> ...
> 
> perhaps the client is a programmer who wants to debug
> "GET /" -> dumps everything
> 
> or just for monitoring (using wget -q0 -)
> "GET /?mode=forewer" -> does it forewer
> 
> i don't intend to use something like soap xml or html. just plaintext  
> over http.
> your thoughts?

Should be pretty to easy to implement if you cannibalize the http_info
plugin. Use exactly one "\t" as column separator and make sure that "\t"
are not used within the columns (I don't know if it is possible but if,
I would replace them with " ") and any reader can easily reformat it.

Does it hurt you if one sends a minimalistic http-header (with
text/plain as content-type and the empty line before the body)?
Then you can actually use a browser to read it.

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services