[csw-maintainers] nginx + passenger

[Jesse Reynolds]

On 16 August 2011 21:38, Dagobert Michelsen <dam at opencsw.org> wrote:

> Hi Jon,
>
> Am 16.08.2011 um 14:05 schrieb Jonathan Craig:
> > On Tue, Aug 16, 2011 at 4:49 AM, Dagobert Michelsen <dam at opencsw.org>
> wrote:
> >>
> >> I suggest adding alternatives with nginx as base package and
> nginx_passenger as
> >> add-on which takes precedence on the binary and depends on ruby etc.
> >
> > So we would have a package that provides function set 'A' as a base
> > package and then create a package that provides function set 'A' + 'B'
> > and uses alternatives to take precedence.  The benefit being that the
> > base package avoids a dependency on ruby?
>
> Yes.
>
> > Would it be better to roll
> > the 'a' + 'b' functionality into the base and create a shell package
> > that appends the ruby dependency but doesn't actually make a binary
> > change.
>
> That will probably not work as the binaries in the base package link
> to dependent libraries not installed. It would also complicate
> checking if dependencies are entered correctly.
>

I'd need to test this, but I think if ruby were not installed, and you ran
up this nginx+passenger binary, it would run fine unless you put passenger
config items into it and expected it to work. That is, I think all the
passenger module provides to nginx is a means to spawn ruby VMs running the
ruby side of the passenger stuff, and to subsequently hand requests to it's
child ruby processes. nginx with passenger support does not link to any new
libraries.


>
> > The benefit of this would be just one actual binary to build
> > and troubleshoot.
>
> True, but for the above reason it usually doesn't work this way.
>

If it were to work this way in this case, is it a good idea to go down this
path?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opencsw.org/pipermail/maintainers/attachments/20110816/60d2dfc7/attachment.html>