[cups.development] Re: IPP Message file parsing

[Jerome Alet]

Hi,

On Sat, Nov 06, 2004 at 06:39:34AM -0500, Michael Sweet wrote:
> Jerome Alet wrote:
> >
> >I've just written an IPP message parser (RFC 2565) which seems to
> >mostly work.
> 
> Actually, you want to use 2910/11 and collections defined in 3382.
> 
> 2565 is obsolete.

ouch !

well, at least it worked with the thousands CUPS message files 
I've tested.

I'll modify it according to the new RFCs though

> >Now I'd want to, from a CUPS backend, access to the IPP message files
> >which are usually in the /var/spool/cups directory
> >
> >But, CUPS doesn't make the exact file name available to backends,
> >perhaps expecting that all backends which want to dialog with CUPS
> >will use the CUPS API, which of course is not the case...
> 
> That is a supported way of doing it; if you want to bypass CUPS, you
> are free to do so, it just isn't supported and may not work in
> future releases since we reserve the right to change how the spool
> directory works at any time.  You aren't using a public interface.

I understand.

> >So my question is :
> >
> >Can I assume that the filename will always be of the form :
> >
> >c00000
> > ...
> 
> No, you can't.  Right now, that is the case.  Future versions of
> CUPS may change the organization.

That was the sense of my question. I meant with actual versions
of CUPS.

> >or is this a configurable setting (especially in the case you want to
> >keep more than 100000 jobs) ?
> 
> Currently it is "c" + a minimum of 5 digits; if the job ID is <
> 5 digits long, it is 0-filled.

Thanks !

Should I file a feature request to have CUPS be a bit more verbose
(with its backends) in the future ? This would probably simplify 
many software, mine included.

Thanks again for your time

bye

Jerome Alet