Update: March 31st: The Lab hopes to have a fix deployed to all three RC channels on Wednesday, April 1st. In terms of the problems related to disabling the HTTP Inventory option, which this fix does not address (see BUG-8917), the Lab notes that going forward, users should keep this option enabled, otherwise issues of load failure will occur. As such, it is anticipated this option will be hidden from general view within the viewer at some point in the future.
The server-side deployment to the three release candidate channels (Bluesteel, LeTigre and Magnum) during week #13 included updates focused on reducing instances of inventory loss, and also included some server-side code clean-up. Unfortunately it also brought with it the potential to create a few inventory-related issues.
The problems are reported in BUG-8877, and have the potential to affect anyone running a version of a viewer that does not have the recent AIS v3 updates (e.g. the current release of Firestorm and, I believe, Singularity), and / or any viewer with or without AIS v3 updates which is running with HTTP Inventory disabled. However, they will only occur when you are actually located on an RC channel region, and then only in situations described below.
You can ascertain whether or not you are on an RC region via the viewer’s Help > About floater.

There are two problems which are being encountered:
- If you empty Trash and relog when using a viewer without the AIS v3 code updates (e.g. the current release of Firestorm), the purged items will reappear in Trash the next time you log-in to SL. This will not happen if you are running a viewer with the AIS v3 updates – your Trash will purge and remain empty, as expected
- If you are running the viewer with HTTP Inventory disabled, and clear cache, your inventory will not fetch as long as you remain on an RC channel region, leaving you a cloud (see below). This will happen regardless of whether you are running a viewer with or without the AIS v3 code updates

Until such time as the server-side code has been updated, these issues can be overcome / avoided by:
- Moving to any non-RC region to purge Trash properly without items returning following a relog
- Re-enabling HTTP inventory in your viewer (CTRL-ALT-Q to display the Develop menu, if required, and then checking HTTP Inventory), and then relogging to overcome issues of inventory fetching following a cache clearance when on an RC region.
The JIRA reporting the issues has been imported by the Lab for an immediate fix. This probably means – subject to confirmation from the Lab – that the code currently on the RC channels will not be promoted to the Main (SLS) channel on Tuesday, March 31st, and that a fix will (hopefully) be deployed to the RC channels on Wednesday April 1st. I’ll have an update in my usual SL project update reports in due course.