[Spacewalk-list] deleting packages

[Sandro "red" Mathys]

On Tue, Nov 23, 2010 at 12:07, Colin Coe <colin.coe at gmail.com> wrote:
> On Tue, Nov 23, 2010 at 4:07 PM, Sandro "red" Mathys
> <red at fedoraproject.org> wrote:
>> On 11/22/2010 11:41 PM, Colin Coe wrote:
>>>
>>> On Tue, Nov 23, 2010 at 2:57 AM, Marcus Moeller<mail at marcus-moeller.de>
>>>  wrote:
>>>>
>>>> Dear Jan.
>>>>
>>>>>>> On you Spacewalk, go to
>>>>>>>
>>>>>>>
>>>>>>>  https://FQDN/rhn/apidoc/handlers/PackagesHandler.jsp#removePackage
>>>>>>>
>>>>>>> Hope this helps,
>>>>>>
>>>>>> You still have to hack orphan package IDs out of the database directly
>>>>>> which is quite annoying.
>>>>>
>>>>> Obviously, patches welcome.
>>>>
>>>> Please note that not all of us are coders. What I can do is to report
>>>> bugs and to help improving the product that way (as I always did in
>>>> the past).
>>>>
>>>> If there will be any form of interest in setting up a QA process I am
>>>> also willed to help.
>>>>
>>>> --
>>>> Greets
>>>> Marcus
>>>>
>>>
>>> Hi all
>>>
>>> I should be able to come up with a 'packages.findOrphans' API call.
>>> It would only be able to be run by a Satellite admin.
>>>
>>> Jan, would that be acceptable from a security/maintenance perspective?
>>>
>>> Marcus, would that help solve the problem in the original post?
>>>
>>> CC
>>>
>>
>> Hi Colin,
>>
>> How would that API call be different from
>> channel.software.listPackagesWithoutChannel?
>>
>> Actually, I think it would help already, if listPackagesWithoutChannel would
>> have some (new) limiting parameters to only select a smaller chunk of
>> packages and thereby work around the 502 problem.
>>
>> -- red
>>
>
> heh, should have done more homework, shouldn't I  :)
>
> I'm wondering if channel.software.listPackageIDsWithoutChannel would
> get around the problem by only returning package IDs?  As the amount
> of data per package would be greatly reduced this may help.
>
> I think the limit idea is a good one.  I guess you mean something like
> channel.software.listPackagesWithoutChannel(sessionKey, 500) to return
> the first 500 packages not in a channel.
>
> It'd be interesting if we could save state so that the first time you
> call channel.software.listPackagesWithoutChannel(sessionKey, 500) in a
> session you get the first 500 packages and subsequent times you get
> the next 500 and so on until the session expires, auth.logoff is
> called, or the last of the packages without a channel are retrieved.

I'd suggest to make it more flexible, i.e. either
listPackagesWithoutChannel(sKey, startOfRange, endOfRange) or lPWC(sK,
startOfRange, rangeSize/numberOfPackages) but both should work for the
case discussed in this thread.

-- red