>>>>> On Sat, 26 Feb 2011 22:55:01 +0100, Magnus Fromreide
>>>>> <[email protected]> said:
MF> I suppose I am still confused by all this so could you please explain
MF> what the relations model?
Sorry for the delay. I took a trip north for work and was away from
mail for a few days.
MF> The third one is somewhat odd:
MF> * If file f.c provides feature X then there is an implementation
MF> of it in file f.c
MF> This only leaves the issue of all features that no file
MF> implements, where do they come from?
I'm not entirely sure what you're asking about. I *think* you're asking
that if a feature is 'required' but nothing implements it, what happens?
Remember that the feature system is really only a marking system and
doesn't actually tie directly to code; only to the creation of ifdefs.
There is no legal binding between code names and feature names. Thus,
this is possible:
netsnmp_feature_provide(bogusname)
#ifdef NETSNMP_FEATURE_HAS_BOGUSNAME
implement_something_entirely_different(){ }
#endif
So, as you can see the name is completely arbitrary.
Also, it allows for this:
netsnmp_feature_require(doesnt_really_exist)
It won't currently complain that you've required a feature that isn't
defined anywhere else. This is actually on my list to fix, but
currently isn't handled. It should, at least, spit a warning or an
error if you do something like that. But again, the real error comes at
code-compile time and whether or not it compiles or not.
MF> The final one is the really hard one:
MF> * If feature X is a child of feature Y then they are connected
MF> how?
That's actually documented on the wiki page:
http://www.net-snmp.org/wiki/index.php/Feature_Marking_and_Selection#Parent.2FChild_Features
Specifically, it says this:
1. if the child is required:
* Will turn on the child
* Will not turn on the parent
2. if the parent is required:
* Will turn on the child
* Will turn on the parent
MF> Creates a dot graph description of the feature relations below a
MF> given path
I thought about doing that. The fact that your diagram is so big shows
me I probably shouldn't!
Right now the local/minimalist/feature-makedocs scripts produces a .org
file that you may find useful, and I hope to update it to produce a wiki
format later today.
--
Wes Hardaker
Please mail all replies to [email protected]
------------------------------------------------------------------------------
What You Don't Know About Data Connectivity CAN Hurt You
This paper provides an overview of data connectivity, details
its effect on application quality, and explores various alternative
solutions. http://p.sf.net/sfu/progress-d2d
_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
