[Nagiosplug-devel] Nagios plugins CPAN-like archive

[Andreas Ericsson]

sean finney wrote:
> hi,
> 
> On Mon, Jul 11, 2005 at 11:10:35PM +0100, Ton Voon wrote:
> 
>>No opinions about a Nagios plugins CPAN-like area? I'm interested to  
>>hear what people think. Please respond privately if you prefer.
> 
> 
> i think providing an easier way for people to submit and review
> plugins is a good idea.  i do have some concerns though:
> 
> my main concern is with qa-related topics.  sure, there's the NIH
> syndrome you mentioned earlier (which i think is relevant), but i'm
> more worried about the overall lack of control over quality,


This shouldn't be a problem, considering the fact that the code quality 
of todays official plugins isn't exactly stellar.


> as well
> as pollution of our namespace/reputation.
> 

The official plugins doesn't HAVE a namespace, unless you count the 
"my_" prefix, which is sort of restricted for personal use by general 
agreement. I like the np_ you added in front of np_connect() though, so 
that could be the seed to one.

If you're worried about polluting official namespaces then that's 
already down the drain (the 6/3 rule is broken by the strnl(), strscpy() 
and whatnot in utils.c, which conflicts nicely with ANSI, POSIX, SUS and 
BSD). Fortunately, those functions can be done away with fairly easily 
and any linker that requires a 6/3 uniqueness in symbol names won't 
compile anything anywhere anyway (now adays, that is).

Whatever you do, stay away from the _t suffix (POSIX) and the __ prefix 
or you'll run into trouble sooner or later. The __ prefix, for those who 
don't know, is reserved for library use. Symbols containing it aren't 
supposed to be visible globally but not all linkers have heard of weak 
aliases, so you're bound to run into trouble sooner or later.


> i think such concerns could be addressed if we had some way of providing
> some kind of nagiosplug "status" or "rating", where it was made clear
> whether or not it was reviewed by a member of the team, whether it was
> being considered for inclusion,


Bad idea, for (at least) two reasons;
1) Plugins in contrib aren't evaluated, checked or any such thing by 
official developers today, so it'd be quite a miracle if they (that is, 
you) started doing so just because you had to download them in a special 
package as opposed to getting them "for free" when checking out the CVS.

2) If some contributor loses commit access to plugins they've originally 
written you'll end up with several forks right away.


> et c.  if such a system existed, i think
> it would make a better bridge for plugins to make their way into the
> project, and would overall be a good thing.  this would also obsolete
> the need for a "contrib" section.
> 
> otherwise, i'd be very hesitant for moving in this direction, and am
> not sure what it would gain us that couldn't be simply done by linking
> to the nagiosexchange site where a similar collection system for plugins
> already seems to exist.
> 

I vote link to nagiosexchange. It's a wiki, so plugin authors can put 
links to whatever they want fairly much wherever they want.

Just make it clear that you frown upon people who add 
nagiosplug-help at lists.sourceforge.net to the help output of their 
plugins, or in any way claim to be affiliated with the official distro 
when they're not.

-- 
Andreas Ericsson                   andreas.ericsson at op5.se
OP5 AB                             www.op5.se
Lead Developer