On Mon, 2018-02-12 at 09:50 -0500, Patrick Shanahan wrote:
* Robby Engelmann <robby.engelmann@igfs-ev.de> [02-12-18 09:25]:
On Montag, 12. Februar 2018 12:04:32 CET Dominique Leuenberger / DimStar wrote:
I know of at leats one user that regularly tests in parallel to openQA, before things 'hit the shelves' - and he did report issues in the past (which we used to block snapshots).
Yes, and that very user was (for whatever reason) not bitten by this issue the time it appeared for most of the users, but one snapshot later. This time the clearing of the cache helped as for all others.
However, after the updated Mesa was removed from the update channel, I was hit again by this issue (or another one), but then clearing the cache did not help and I made a fresh install (in order to test the fresh storage-stack).
not that it makes any diff now, but I also got "hit" again when Mesa reverted, but removing the cache items again solved it and the solution has persisted since.
There were no Mesa reverts. What did happen is a 'release number decrease' as the same code built in :Update is '1 ahead of non-Update' (which is good, as this helps us ensure Update is 'higher'); once the update is merged though into the mainline tree, the :Update is emptied, and the 'count' reset to what is in mainline. It was 100% identical sources (but still different build id, which is what makes Mesa discard the cache and Qt falling on the nose for not recompiline the Gl shaders) Cheers Dominique