No limit on job rest call

[robert]

On the REST call /asset/$asset_id/joblist is no limit on items that it return. In most cases this will not be a problem. But it can happen that assets will build up a lot of old jobs which might cause problems on the output this REST call can generate (memory, long time the REST call will return output etc).

Scope: 1.x and up.

[Frans]

Not only a numerical limit, but also on job status in order to filter not completed or running jobs.
SURFmedia asks for /asset/{id}/joblist on every item.html, and it is looking for not-finished trancodings or transcodings in progress. Finished transcodings and finished analysis could be omitted by e.g. /joblist?hide_finished=true or /joblist?running=true or so.

[robert]

Frans,

Is it ok to use CQL here? I have this terrible feeling that if we will be using old-fashion syntax like created=date|date and status[]=INPROGRESS or status_not[]=INPROGRESS we will be adding so many new parameters and options to this rest call, while using CQL it should be a lot easier and lot more flexible. Ill provide for a extended manual on this rest call how to search on each parameter with CQL.

So ok to implement with CQL parameter or with normal parameters per field/option?

[Frans]

Robert, Using CQL here is good and flexible. So go ahead!
The extended CQL manual is also welcome, so go ahead with this too ;-)

[Michiel.Schok]

A Bug and an Question:

Bug: when limit is set below the total amount of jobs, 'item_count_total' doesn't reflect the total, but is equal to 'item_count'.

We have an asset with three (3) jobs:

  <header>
    <item_count>3</item_count>
    <item_count_total>3</item_count_total>
    <item_offset>0</item_offset>
...
    <request_uri>[GET] asset/Pf1HVZWuKUPFZh9zKqit8DDe/joblist?user_id=SURFguest.Rijnmond</request_uri>
    <version>2.2.0.727:2453fceb2426</version>
    <request_query_count>2</request_query_count>
  </header>

When we only request 2 of them (we want to introduce pagination ;-) )

  <header>
    <item_count>2</item_count>
    <item_count_total>2</item_count_total>
    <item_offset>0</item_offset>
...
    <request_uri>[GET] asset/Pf1HVZWuKUPFZh9zKqit8DDe/joblist?user_id=SURFguest.Rijnmond&amp;limit=2</request_uri>
...
  </header>

Question: /admin/mediamosa/browse/restcall/asset-%24asset_id-joblist/GET
shows some CQL examples. Are there any other fields that can be used? The last two examples: do they parse <error_description>?

[forgacs]

Fixed: item_count_total.

Fields that can be used:

job_id
app_id
mediafile_id
owner_id
status
progress
priority
job_type
started
finished
create_still
no_hint
still_parameters
error_description
created
changed

"create_still" and "no_hint": The response "error_description" is the "error_description" field of the job without any modification.

Robert: Please, review the changes.

[Michiel.Schok]

Curious:
when I issue a

[GET] asset/P2WQuIonDsIuTdePEbmNdmKV/joblist?user_id=SURFgroepen.michiel71

I dont get a message saying "you are not allowed to view the joblist" but rather a

    <request_result>success</request_result>
    <request_result_description>The query returned an empty result set (i.e. zero rows).</request_result_description>

The same request using the owner_id of the asset as user_id returns the joblist.

But the rest is working as advertised.
Checked using limit=5, limit=5&offset=10, cql=(status==finished), cql=(job_type=analyse)

[robert]

About your "you are not allowed to view the joblist";

Hmm hmm. That is some curious question. Do you need to be owner of the asset to see the jobs? You don't need to be owner of the asset to see the asset in normal asset call. Think you need to compare the search on the joblist as the search on assets. If no assets where found (because of access) then you still retrieve a empty result message.

I do see that when supplying the is_app_admin = TRUE parameter you can retrieve any job(s) of any asset of any app. Looking at the code in 1.7 also, its been designed like this. Maybe need some work as well?

[robert]

See #356 and #357