[Nitro] Simpler configuration method?

transfire at gmail.com transfire at gmail.com
Thu Jan 25 19:28:42 EST 2007



On Jan 25, 5:42 pm, "George Moschovitis" <george.moschovi... at gmail.com>
wrote:
> Tom,
>
> On 1/25/07, George Moschovitis <george.moschovi... at gmail.com> wrote:
>
>
>
> > Imagine if you will a mod_nitro plugin for apache, rather than create
> > > multiple execution spaces for each nitro app, single space is use to
> > > improve preformace. In such a case global variables will be shared
> > > between them and everything goes to pot.
>
> > hmm, this is a killer. But to tell you truth, I find it quite improbable
> > that a single ruby interpretter will run many Nitro applications.
> > Anyway, you have a (strong) point.now that I am thinking about it, the current implementation (ie
> Server.root_dir instead of $server_root_dir) will not work in the scenario
> you propose as well. Server.root_dir is in fact a (less flexible as
> described earlier) global variable. Am I missing something here?

Nope. You hit the nail on the head. Which is why I brogh it up -- if
youre thinking about refactoring this, I encourage you to do so in the
direction of a Server instance. It will take some careful thought I'm
sure --globals tend to be very easy since they are univerally
accessable. Using instances well on the other hand requires that the
object model be very tight. I would recommend reading up on IOC
(inverson of control) that more than anything could probably have a
fantastic impact on this design. 

T.



More information about the Nitro-general mailing list