See merge request nexedi/slapos.core!524

See merge request nexedi/slapos.core!530

See merge request nexedi/slapos.core!528

See merge request nexedi/slapos.core!526

See merge request nexedi/slapos.core!527

    if uid is empty it returns whole catalog

See merge request nexedi/slapos.core!529

   In short, use PUT rather them GET to request instance

    Use response from putAttachment to get the relative URL

   Call catalog on the uids for compute properly the list of objects to list (pagination + sort).

    Remarrk: This change isn't ideal and introduce an extra catalog call, but this only adds few microseconds (>10ms) to the return.

   - Fixup current Organisation on Compute Node UI (missing field)
   - Extend the query to fetch as much project and organisations as possible
   - Sort alphabetically, since it is more user_friendly
   - Keep single field (not current/Future) for each project, and pre-select the default

   - Fixup current Organisation on Compute Node UI (missing field)
   - Extend the query to fetch as much project and organisations as possible
   - Sort alphabetically, since it is more user_friendly
   - Keep single field (not current/Future) for each project, and pre-select the default

   This was prevent request form to restrict the software type to the proper one.

See merge request nexedi/slapos.core!525

See merge request nexedi/slapos.core!523

  Since priority is 1 the activies can be called out of order. So calculateEtags can be called before all objects be indexed. This can lead to a intermediate (or old) value, requiring am extra call to the _activateFillComputeNodeInformationCache.

  This is somehow acceptable since the code should expect an ordering of activities execution anyway.

   This prevents to be delayed due excessive activities created by simulation

See merge request nexedi/slapos.core!520

   Dont fail if the Software Release couldnt be found, just stop (display error explanation) and allow the user to return where he came from.

  The string comes translated.

   This wasnt a proper update, and it was accidently commited by mistake on the MR.

See merge request nexedi/slapos.core!522

   The proper way is always wait for notification message after click proceed, since the spinning icon can flick between before the post and reload the page (probably because notifySubmitted).

   This tiny flickering can cause confuse the wait_for_content macro since it is called too soon.

   Mostly to translate Open allocation state (causing test to fail)

See merge request nexedi/slapos.core!518

   The selenium dont assert well with \n and code, so play safe and just assert line by line.

   This ensure data is saved before render a new form.