sinatra is not working with passenger 2.2.0 on nginx?

[hui]

I am trying to run sinatra-rubygems (http://github.com/jnewland/
sinatra-rubygems/tree/master) on passenger 2.2.0, but all requests
returns 404 not found.

"rackup config.ru" (http://localhost:9292) is OK, and error.log for
nginx shows:
'127.0.0.1 - - [17/Apr/2009 16:45:41] "GET /Users/hui/Sites/sinatra-
rubygems/public " 404 18 0.0008'

it seems for sinatra all "@request.path_info" (==PATH_INFO?) is set to
"root", which in my case is "/Users/hui/Sites/sinatra-rubygems/
public", and leads to raising a 404 error.

my nginx config:

server {
listen 80;
server_name gems.local;
root /Users/hui/Sites/sinatra-rubygems/public;
passenger_enabled on;
}

config.ru for sinatra-rubygems

LIB_PATH = File.expand_path(File.join(File.dirname(__FILE__), 'lib'))
$:.unshift LIB_PATH
require 'rack_rubygems'
use GemsAndRdocs, :urls => ['/cache', '/doc'], :root => Gem.dir
use Rack::Compress
run RackRubygems.new


anyone has some idea about this?
thx.

[Tim Carey-Smith]

I found this problem too.
I wrote a dodgy Rack middleware which will "fix" the problem.

https://gist.github.com/e4337fdf7722a9e44e82

I'm going to try and track down the real cause now.

Ciao,
Tim

[Tim Carey-Smith]

I have traced the problem to before it gets to ruby.

These lines in the nginx content handler seem to be doing the wrong
thing, I think.

http://github.com/FooBarWidget/passenger/blob/master/ext/nginx/ContentHandler.c#L289
http://github.com/FooBarWidget/passenger/blob/master/ext/nginx/ContentHandler.c#L414-416

I am looking into how to convert the REQUEST_URI into the PATH_INFO
more correctly than my simple regex.

Ciao,
Tim

[dylan]

I think this one is ok:

env["PATH_INFO"] = env["REQUEST_URI"].split('?', 2).first

from http://github.com/chneukirchen/rack/blob/663abfce9b71ab273c8adc0e76c233671bc43e1d/lib/rack/handler/scgi.rb#L30

On Apr 18, 10:41 pm, Tim Carey-Smith <g...@spork.in> wrote:
> I have traced the problem to before it gets to ruby.
>
> These lines in the nginx content handler seem to be doing the wrong  
> thing, I think.
>

> http://github.com/FooBarWidget/passenger/blob/master/ext/nginx/Conten...http://github.com/FooBarWidget/passenger/blob/master/ext/nginx/Conten...

[Tim Carey-Smith]

I have fixed the problem on my fork.

http://github.com/halorgium/passenger/commit/b8b360ea1d960e965fe7a7fec3df9a36aa04e13b
http://github.com/halorgium/passenger/commit/6d64144b1c57e5b7788537fe07dca1f3e9e7c388

Two commits cause I almost failed miserably :)

Can someone check that I'm not doing something insanely stupid?

Thanks Dylan for the split line.

Ciao,
Tim

[Hongli Lai]

On Sun, Apr 19, 2009 at 1:27 AM, Tim Carey-Smith <g...@spork.in> wrote:
>
> I have fixed the problem on my fork.
>
> http://github.com/halorgium/passenger/commit/b8b360ea1d960e965fe7a7fec3df9a36aa04e13b
> http://github.com/halorgium/passenger/commit/6d64144b1c57e5b7788537fe07dca1f3e9e7c388
>
> Two commits cause I almost failed miserably :)
>
> Can someone check that I'm not doing something insanely stupid?
>
> Thanks Dylan for the split line.

Thanks.

Can anybody confirm whether this change doesn't break Sinatra on Apache?

--
Phusion | The Computer Science Company

Web: http://www.phusion.nl/
E-mail: in...@phusion.nl
Chamber of commerce no: 08173483 (The Netherlands)

[Tim Carey-Smith]

On 19/04/2009, at 9:25 PM, Hongli Lai wrote:

>
> On Sun, Apr 19, 2009 at 1:27 AM, Tim Carey-Smith <g...@spork.in> wrote:
>>
>> I have fixed the problem on my fork.
>>
>> http://github.com/halorgium/passenger/commit/b8b360ea1d960e965fe7a7fec3df9a36aa04e13b
>> http://github.com/halorgium/passenger/commit/6d64144b1c57e5b7788537fe07dca1f3e9e7c388
>>
>> Two commits cause I almost failed miserably :)
>>
>> Can someone check that I'm not doing something insanely stupid?
>>
>> Thanks Dylan for the split line.
>
> Thanks.
>
> Can anybody confirm whether this change doesn't break Sinatra on
> Apache?

The only thing which this changes is that it will overwrite the
PATH_INFO which apache provides.
Maybe it should change from:

> headers["PATH_INFO"] = headers["REQUEST_URI"].split("?", 2).first

to:

> headers["PATH_INFO"] ||= headers["REQUEST_URI"].split("?", 2).first

So it'll keep the existing value.

Ciao,
Tim

[Eric Watson]

On Apr 19, 5:20 am, Tim Carey-Smith <g...@spork.in> wrote:
> On 19/04/2009, at 9:25 PM, Hongli Lai wrote:
>
>
>
>
>
> > On Sun, Apr 19, 2009 at 1:27 AM, Tim Carey-Smith <g...@spork.in> wrote:
>
> >> I have fixed the problem on my fork.
>

> >>http://github.com/halorgium/passenger/commit/b8b360ea1d960e965fe7a7fe...
> >>http://github.com/halorgium/passenger/commit/6d64144b1c57e5b7788537fe...

>
> >> Two commits cause I almost failed miserably :)
>
> >> Can someone check that I'm not doing something insanely stupid?
>
> >> Thanks Dylan for the split line.
>
> > Thanks.
>
> > Can anybody confirm whether this change doesn't break Sinatra on  
> > Apache?
>
> The only thing which this changes is that it will overwrite the  
> PATH_INFO which apache provides.
> Maybe it should change from:
>
> > headers["PATH_INFO"] = headers["REQUEST_URI"].split("?", 2).first
>
> to:
>
> > headers["PATH_INFO"] ||= headers["REQUEST_URI"].split("?", 2).first
>
> So it'll keep the existing value.
>
> Ciao,
> Tim

Thanks for tracking this down and fixing it, Tim. I had the same
problem.

Eric

[gaiusparx]

Hi Tim Is your fork based on 2.2.1 or 2.2.0?

Hi Phusion, is this patch merged back to the git://github.com/FooBarWidget/passenger.git
already?

Thanks

On Apr 19, 7:27 am, Tim Carey-Smith <g...@spork.in> wrote:
> I have fixed the problem on my fork.
>

> http://github.com/halorgium/passenger/commit/b8b360ea1d960e965fe7a7fe...http://github.com/halorgium/passenger/commit/6d64144b1c57e5b7788537fe...

>
> Two commits cause I almost failed miserably :)
>
> Can someone check that I'm not doing something insanely stupid?
>
> Thanks Dylan for the split line.
>
> Ciao,
> Tim
>
> On 19/04/2009, at 5:08 AM, dylan wrote:
>
>
>
>
>
> > I think this one is ok:
>
> >   env["PATH_INFO"] = env["REQUEST_URI"].split('?', 2).first
>

> > fromhttp://github.com/chneukirchen/rack/blob/663abfce9b71ab273c8adc0e76c2...

>
> > On Apr 18, 10:41 pm, Tim Carey-Smith <g...@spork.in> wrote:
> >> I have traced the problem to before it gets to ruby.
>
> >> These lines in the nginx content handler seem to be doing the wrong
> >> thing, I think.
>
> >>http://github.com/FooBarWidget/passenger/blob/master/ext/nginx/Conten...

[Tim Carey-Smith]

On 23/04/2009, at 11:00 AM, gaiusparx wrote:

>
> Hi Tim Is your fork based on 2.2.1 or 2.2.0?

This is based off of 2.2.1.

Thanks,
Tim

[Hongli Lai]

On Thu, Apr 23, 2009 at 1:00 AM, gaiusparx <xen...@gmail.com> wrote:
>
> Hi Tim Is your fork based on 2.2.1 or 2.2.0?
>
> Hi Phusion, is this patch merged back to the git://github.com/FooBarWidget/passenger.git
> already?

Not yet, we haven't had the time to review it yet but it's on the top
of our todo list.

[Hongli Lai]

The problem has been fixed in git now.

[Eric Watson]

On Fri, Apr 24, 2009 at 11:48 AM, Hongli Lai <hon...@phusion.nl> wrote:

The problem has been fixed in git now.

Thanks, Hongli. I reinstalled passenger 2.2.1 from your github repo, rebuilt my arch linux package of nginx with the passenger module, and it works great.

Thanks again to Tim, too.

Eric