On 10/1/10 6:59 AM, "no-re...@cfengine.com" <no-re...@cfengine.com> wrote:
> Well... I think that is a good idea basicly in the best of worlds but
> sometimes things need to happen in a certain order.
While it requires some change of traditional thinking, the confusion often
comes when we think things "need to happen in a certain order, AND in one
pass" -- vs accepting that, with the correct class definitions, cfengine can
do things in the right order over multiple passes (convergence). More than
the order itself, it's often the desire to see immediate results (instant
gratification) that we must let go of in cfengine.
> Just because it is a state description doesn't mean that is should be
> incomprehensable and/or hard to read. With CFE, simple promices becomes easy
> to read but the more complex ones becomes a mess that makes no sens. I have
> seen collegues solve some problems by just using a file promices to get a
> script in place and a command promise to execute it... and that is just ugly
> (or is this the intended way to do it?).
That depends on what you push into the script, IMCO. If you push lots of
little actions cfe can easily manage, then you are losing some benefit. If
you push application customization or other things into an install script or
similar, then it might make perfect sense. I guess I don't look at
solutions as so "right" or "wrong" (I thought scaling Web 2.0 taught us all
to accept a little "wrong" in tech solutions?) -- if it works, and testing
reveals it's reliable... I'm pretty happy.
That said, maybe this is just an interim state. You can use a copy
statement and a script, and continuously fold more and more of your external
scripts into cfengine policy over time.
> If I should be able to think at the end state first then I should be able to
> write the promices as high level and then write library functions that dealt
> with the details about how to get there? But with CFE I can't think of any way
> how to do this. I I don't mean that I should use black magic to get there.
"Any sufficiently advanced technology is indistinguishable from magic."
--- Arthur C. Clarke
:)
> I belive that CFE should benefit of a good jolt of pragmatic ideas, a more
> sane syntax and better documentation.
FWIW, I felt the same pain while coming up to speed on cf2 originally. The
docs got better over time, as did my own understanding. cf3 is still fairly
new. I suspect the documentation will continue to grow and improve.
The syntax is not something I love, honestly, but to imply it is insane
would need examples of inconsistency. I won't argue there is a steep
learning curve, but Mark has mentioned that is because there are a lot of
use cases being addressed that can't be simplified away. If you find
inconsistent behavior which causes confusion, it might be a bug.
_______________________________________________
Help-cfengine mailing list
Help-cfengine@cfengine.org
https://cfengine.org/mailman/listinfo/help-cfengine
