Re: it scares me a bit...

[Jan-Henrik Haukeland]

I get you! And just to make it clear, feedback like yours and Michael's 
are welcome and we do read them carefully. You both have reasonable 
arguments for sharpening the focus of monit and not allow rand 
extensions, but we are not in disagreement about this!

I expect that your concerns are a result of the latest feature request 
and the decision to implement it. But as I said before, nothing in this 
request is new. If you check the monit code and documentation you can 
see that monit *already* has url monitoring in that you can request a 
document from a http server and even test the checksum of the content. 
This is useful and used by many. The only new thing in this request is 
the ability to do a regexp search on the content. This is hardly a 
feature creep and adding bloat to monit! On the contrary, this only 
sharpen the monitoring abilities of monit. And as Martin implied, as 
long as we stay inside our problem domain which is monitoring and keep 
it simple, then there should be no problems. Agree?

If you reacted to my flippant statement about the "need to continuously 
add features to monit to stay alive" then don't take that to seriously. 
I do not mean that we should end up with the ability to play MP3s and 
read mail with monit. However it *is* true that if no-one is working on 
the code, the project is pretty much dead and replaced with another. 
There are plenty of examples on this, take for instance Mosaic and the 
NCSA web-server. Like Apache, which has implemented the HTTP/1.1 
protocol a long time ago but still are worked on, I think and hope that 
we will continue to work on monit and add new features *within* the 
monitoring problem domain - without having to be suspected of adding 
bloat to the code.

Michael had a good idea which we already use, that is, to utilize 
m/monit for specialized stuff that does not belong in monit. BTW a new 
version of m/monit should be ready 1. quarter next year with a couple 
of interesting stuff for those that are serious about their monitoring 
business in large :-)


On Dec 4, 2004, at 18:18, Vlada Macek wrote:

> Hi guys,
>
> I just like to tell you that Michael Shigorin is not the only one here
> with such concerns. There were some intentions to use monit as a core
> service started/stopper, replacement for init. Although I don't use it
> this way, I liked to think that I use a tool with such big ambitions...
>
> Now I'm afraid we are slighly diverging from the spirit of good unix
> tool. IIRC, such tool is actually not designed to be the swiss army
> knife -- to open you every can. Unix is the collection of simple tools
> that wisely interact to fulfil whatever the user needs. Every tool does
> only one thing and does it perfectly, being modifiable by reasonable
> options.
>
> I do not agree the project must be added features all-the-time to be
> considered alive. There will be bugs to fix or existing design
> imperfections to refine. I would eventually welcome the statement on 
> the
> project page "We, the developers, think the work is done. This project
> is feature-full and there are no known bugs. The development is closed,
> until you find one. Happy using."
>
> I consider such moment to be great success, not the death. There are so
> many other goals in the world, that can be achieved...
>
> With this post I do not want to touch anyone, I just beg the
> responsible. Please, always think hard before you accept new ideas, 
> have
> no mercy. Get the lot of user feedback on the topic first. You may make
> one or two users happy with the feature, but things get more difficult
> for the rest of the community from that moment! Also newscomers will
> have it harder to read the docs, many more wont manage it...
>
> Simplicity is the key.
>
> I hope someone gets me.
>
> Thanks for monit.