[kronolith] Event notification from another user private calendar

[Luis Felipe Marzagao]

Em 18-02-2016 21:47, Jens Wahnes escreveu:
> On Thu, Feb 18 2016, at 17:18:34 -0500, Michael J Rubinsky wrote:
>
>> Quoting Luis Felipe Marzagao <lfbm.andamentos at gmail.com>:
>>> Em 18/02/2016 18:31, Michael J Rubinsky escreveu:
>>>> Quoting Jens Wahnes <wahnes at uni-koeln.de>:
>>>>> SELECT * FROM kronolith_sharesng WHERE share_name = 0 [pid 18524 on
>>>>> line 321 of "/usr/share/pear/Horde/Db/Adapter/Mysqli.php"]
>>>> This is probably coming from Kronolith_Application::listAlarms().
>>>>
>>>> As an admin user, execute the following from the admin phpshell page,
>>>> using kronolith's scope:
>>>> You should see an array of share_names. If I'm right, one of these will
>>>> be '0' for some unknown reason.
>>> Yep, for me it does. The 23th item:
>>>    [23]=>
>>>    int(0)
> Similar thing here, only that it was the first (index 0) element of the
> array.
>
>> This means that the share in the database doesn't have a share_name for
>> some reason. Can you see which share that is in the kronolith_sharesng
>> table?
> Yes, there was an entry that - beside the share_id - was completely
> "empty", that is share_name was empty, share_owner was NULL and all
> permissions were 0.
>
> I deleted that empty share entry in kronolith_sharesng, removed all
> duplicated entries from horde_alarms and reran the horde-alarms script.
> No duplicate entries were re-created. Yippee.
>
> Now for how that empty kronolith_share was created, I don't really
> know. If I go by the share_id, it must have been created back in the
> time when we ran Horde 3 (or Horde 2 even). But maybe it wasn't created
> that way but got changed to that later on? Hopefully Luis can shed some
> light on that from his case - he seems to have far less users/shares
> than we do.

Yes, my deploy is very small. And my system was also updated from good 
old horde 3 or 2.

The trouble in my side is I'm not locating any entry without (or with 
empty) share_name or share_id.

Take a look here: http://marc.info/?l=kronolith&m=145583830721143&w=2

So I'm guessing Kronolith_Shares::listAllShares, in my case, is 
generating that last array member (0) on the fly.

I'll keep looking into it.

>
> Jens
>
>