On Thu, Apr 6, 2017 at 9:05 PM, <nick@freshcoders.nl> wrote:
Hi,

While testing with Avacado, our team will be generating a zip file. This file will contain files such as screenshots. Which we will store on a server. To link to this server from ResultsDB, we were planning on putting the link in the 'ref_url' field.

In the docs (http://docs.resultsdb20.apiary.io/#reference/0/results) it is stated this field is used for documentation and issues regarding the test. There also is a data field in the ResultsDB which allows custom fields. Which one would suit the zip file better?

Regards, Nick Dekker

Hi Nick,

are you going to run your tasks through Taskotron, or run them elsewhere and just submit results to ResultsDB?

If you're going to use Taskotron, you can save any number of artifacts into ${artifactsdir}, and they will be uploaded to BuildBot server after execution and available publicly. The ref_url can contain a single link, so usually it's the task-created output file (e.g. rpmlint output). If it's not filled out, it points to the overall task stdout (including initialization and everything). If you need to show several artifacts (a task output, a picture, etc) to the user, a good idea is to create a simple html file, link/embed the artifacts there and set it as the ref_url. If you don't care about end-users much (it's not meant for e.g. Fedora packagers to inspect, but it's for your internal consumption only), you don't need to care about ref_url at all, just store the artifact and you can programatically access it (once you receive a new results notification containing ref_url, you can replace the filename at the end of the url with the artifacts filename you want to retrieve).

If you're not going to use Taskotron, you can use ref_url to point to the zip/dir stored on your server. It will be a bit harder to access the task stdout, but that's a problem on our end - we need to make it easier in case people point somewhere else. An alternative is to use a custom field for that, e.g. 'zip_url'. The downside is that general audience will have no clue how to access your artifacts. Both approaches are valid, I'd pick the first one.

Does that help?