Generally improve error handling in ARM deployments
I've just experienced an issue where my deployment got stuck at the "Accepted" state for 45 minutes before I decided to cancel it. The actual cause of the hang remains a mystery, but I can make an educated guess. The resource that caused the hang (I know this because it was a linked deployment deploying only an sql server with a database) was deleted along with the resource group before I tried recreating it. Deleting this resoure probably takes more time than the Azure Portal gives the impression of (as it is with several resource types in Azure). So when I try to redeploy the same resource it gets in a state where it is both deploying and deleting at the same time. But this state is not handled nor communicated to me as a user.
This is just one concrete example, but it is clear given the share amount of other issues here regarding error messages that Azure has a serious problem in that area. It would be much appriciated if work was done on this in favour of releasing new features for a while. New features are cool, but the overall experience deteriorates when the feedback we as users get is so lacking.