So along these lines then this must be a bug in 'pkgsend generate':

'pkgsend generate' will happily turn these pkgmap entries:
1 p none etc/scn/scn_aa_read 0600 root sys
1 p none etc/scn/scn_af_read 0600 root sys

into:
unknown path=etc/scn/scn_aa_read
unknown path=etc/scn/scn_af_read

and still report success. I would have expected an error return here
if porting the pipe entries in a pkgmap is not supported.

Is this a known issue?

Your earlier comment -
>> However, pkgsend still returns an error code which breaks
>> my build unless I ignore error returns which I'd rather not. Is there any 
>> way to suppress this particular error
>> condition? 
> 
> Yes, fix the SVR4 package before running pkgsend generate on it.

But the svr4 package isn't broken...

thanks
Dave


On Jul 25, 2011, at 3:06 PM, Tim Foster wrote:

> Hi Dave,
> 
> On Mon, 2011-07-25 at 14:45 -0700, David Sechrest wrote:
>> pkgsend: generate: ERROR: class action script used in
>> SUNWscnconnmgt-ra: var/opt/sun/xvm/security/jsse/scn-agent/truststore
>> belongs to "preserve" class
>> pkgsend: generate: ERROR: class action script used in
>> SUNWscnconnmgt-ra:
>> var/opt/sun/xvm/persistence/scn-agent/connection.properties belongs to
>> "preserve" class
>> 
>> But it looks like the manifest is still generated correctly and has the 
>> following 2 entries for the above:
>> 
>> file reloc/var/opt/sun/xvm/persistence/scn-agent/connection.properties 
>> group=root mode=0600 owner=root 
>> path=var/opt/sun/xvm/persistence/scn-agent/connection.properties 
>> preserve=true
>> file reloc/var/opt/sun/xvm/security/jsse/scn-agent/truststore group=root 
>> mode=0600 owner=root path=var/opt/sun/xvm/security/jsse/scn-agent/truststore 
>> preserve=true
> 
> Yes, we generate the manifest, but intentionally report an error - we've
> no idea what that 'preserve' class action script actually does: we can't
> assume that a package hasn't modified it to alter the SVR4 contents once
> installed, in which case, the error code is the right thing to do.
> 
> The preserve class itself doesn't appear to have any special meaning, at
> least according to the pkgmap(4) man page - so it's by convention that
> Solaris packages have always used a 'preserve' class in this way.  I do
> wonder whether that tag should have been propagated across into pkgsend
> output though.
> 
> Perhaps I'm being too strict?
> 
>> However, pkgsend still returns an error code which breaks
>> my build unless I ignore error returns which I'd rather not. Is there any 
>> way to suppress this particular error
>> condition? 
> 
> Yes, fix the SVR4 package before running pkgsend generate on it.
> 
>       cheers,
>                       tim
> 
> 

---
David Sechrest                            Phone:  (408) 276-5800 x15800
Systems Management                        Email:  [email protected]
Oracle Corporation


_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to