You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Manoj Kasichainula <ma...@raleigh.ibm.com> on 1999/11/01 23:38:08 UTC

APR pipes can't be timed out?

Bill and I were looking at APR pipes, which are implemented as
ap_file_t. It appears that ap_file_t doesn't have support for
timeouts, so we can't timeout CGIs right now.

I guess the proper solution would be to either add timeouts to
ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
work will pick this up and implement it. Pleeeeeease?

-- 
Manoj Kasichainula - manojk@raleigh.ibm.com
IBM, Apache Development

Re: APR pipes can't be timed out?

Posted by Ben Laurie <be...@algroup.co.uk>.
Manoj Kasichainula wrote:
> 
> On Wed, Nov 03, 1999 at 07:22:28PM +0000, Ben Laurie wrote:
> > Manoj Kasichainula wrote:
> > > I guess the proper solution would be to either add timeouts to
> > > ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
> > > work will pick this up and implement it. Pleeeeeease?
> >
> > It'll have to be ap_pipe_t. It isn't reasonable to expect general files
> > to be timeoutable.
> 
> If it was ap_pipe_t, then we couldn't pass a pipe into any functions
> that take ap_file_t. We'd lose a lot of the flexibility that Unix (and
> even Windows, I think) gives us.
> 
> How about returning APR_ENOTIMPL when trying to set a timeout on a
> file that can't handle it? Yeah, it's icky.

Except it is the wrong error. It should be EINVAL, no?

Cheers,

Ben.

--
http://www.apache-ssl.org/ben.html

"My grandfather once told me that there are two kinds of people: those
who work and those who take the credit. He told me to try to be in the
first group; there was less competition there."
     - Indira Gandhi

Re: APR pipes can't be timed out?

Posted by Ben Laurie <be...@algroup.co.uk>.
Manoj Kasichainula wrote:
> 
> On Wed, Nov 03, 1999 at 07:22:28PM +0000, Ben Laurie wrote:
> > Manoj Kasichainula wrote:
> > > I guess the proper solution would be to either add timeouts to
> > > ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
> > > work will pick this up and implement it. Pleeeeeease?
> >
> > It'll have to be ap_pipe_t. It isn't reasonable to expect general files
> > to be timeoutable.
> 
> If it was ap_pipe_t, then we couldn't pass a pipe into any functions
> that take ap_file_t. We'd lose a lot of the flexibility that Unix (and
> even Windows, I think) gives us.
> 
> How about returning APR_ENOTIMPL when trying to set a timeout on a
> file that can't handle it? Yeah, it's icky.

I can live with that.

Cheers,

Ben.

--
http://www.apache-ssl.org/ben.html

"My grandfather once told me that there are two kinds of people: those
who work and those who take the credit. He told me to try to be in the
first group; there was less competition there."
     - Indira Gandhi

Re: APR pipes can't be timed out?

Posted by Manoj Kasichainula <ma...@io.com>.
On Wed, Nov 03, 1999 at 07:22:28PM +0000, Ben Laurie wrote:
> Manoj Kasichainula wrote:
> > I guess the proper solution would be to either add timeouts to
> > ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
> > work will pick this up and implement it. Pleeeeeease?
> 
> It'll have to be ap_pipe_t. It isn't reasonable to expect general files
> to be timeoutable.

If it was ap_pipe_t, then we couldn't pass a pipe into any functions
that take ap_file_t. We'd lose a lot of the flexibility that Unix (and
even Windows, I think) gives us.

How about returning APR_ENOTIMPL when trying to set a timeout on a
file that can't handle it? Yeah, it's icky.

-- 
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/

Re: APR pipes can't be timed out?

Posted by Ben Laurie <be...@algroup.co.uk>.
Manoj Kasichainula wrote:
> 
> Bill and I were looking at APR pipes, which are implemented as
> ap_file_t. It appears that ap_file_t doesn't have support for
> timeouts, so we can't timeout CGIs right now.
> 
> I guess the proper solution would be to either add timeouts to
> ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
> work will pick this up and implement it. Pleeeeeease?

It'll have to be ap_pipe_t. It isn't reasonable to expect general files
to be timeoutable.

Cheers,

Ben.

--
http://www.apache-ssl.org/ben.html

"My grandfather once told me that there are two kinds of people: those
who work and those who take the credit. He told me to try to be in the
first group; there was less competition there."
     - Indira Gandhi

Re: APR pipes can't be timed out?

Posted by Ryan Bloom <rb...@raleigh.ibm.com>.
It is, Manoj and I started talking about this yesterday, but I had to
leave to have some tests run.  You can expect to see layering in APR
sometime late this week or next week, because both of us are interested in
seeing it happen.  :)  Of course, I don't want to speak for Manoj, but I
think he'd agree about this.

Ryan

On Mon, 8 Nov 1999, Dean Gaudet wrote:

> sorry... didn't mean to be obscure... it might be the right time to move
> layering into APR though.
> 
> Dean
> 
> On Mon, 8 Nov 1999, Manoj Kasichainula wrote:
> 
> > On Thu, Nov 04, 1999 at 08:01:09AM -0800, Dean Gaudet wrote:
> > > iol_pipe ?
> > 
> > What this meant finally clicked in my brain, and this seems like a
> > most excellent solution.
> > 
> > -- 
> > Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> > 
> 

_______________________________________________________________________
Ryan Bloom		rbb@raleigh.ibm.com
4205 S Miami Blvd	
RTP, NC 27709		It's a beautiful sight to see good dancers 
			doing simple steps.  It's a painful sight to
			see beginners doing complicated patterns.	


Re: APR pipes can't be timed out?

Posted by Dean Gaudet <dg...@arctic.org>.
sorry... didn't mean to be obscure... it might be the right time to move
layering into APR though.

Dean

On Mon, 8 Nov 1999, Manoj Kasichainula wrote:

> On Thu, Nov 04, 1999 at 08:01:09AM -0800, Dean Gaudet wrote:
> > iol_pipe ?
> 
> What this meant finally clicked in my brain, and this seems like a
> most excellent solution.
> 
> -- 
> Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> 


Re: APR pipes can't be timed out?

Posted by Manoj Kasichainula <ma...@io.com>.
On Thu, Nov 04, 1999 at 08:01:09AM -0800, Dean Gaudet wrote:
> iol_pipe ?

What this meant finally clicked in my brain, and this seems like a
most excellent solution.

-- 
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/

Re: APR pipes can't be timed out?

Posted by Dean Gaudet <dg...@arctic.org>.
iol_pipe ?

Dean

On Mon, 1 Nov 1999, Manoj Kasichainula wrote:

> Bill and I were looking at APR pipes, which are implemented as
> ap_file_t. It appears that ap_file_t doesn't have support for
> timeouts, so we can't timeout CGIs right now.
> 
> I guess the proper solution would be to either add timeouts to
> ap_file_t or create ap_pipe_t. I'm hoping someone who is doing APR
> work will pick this up and implement it. Pleeeeeease?
> 
> -- 
> Manoj Kasichainula - manojk@raleigh.ibm.com
> IBM, Apache Development
>