Dropbox is shutting down its two apps: Mailbox and Carousel. I’ve been using both apps heavily recently. The most important feature of Mailbox is auto-swipe. Other competitors also have similar features, but they require much more effort to setup, while Mailbox only requires tap and hold. Easy like a piece of cake. Auto-swipe saves me from going through 14k+ emails so far. Carousel saved me lots of iPhone space by uploading my photos to Dropbox. I heard that they will integrate important features of Carousel into Dropbox, so I will not miss Carousel too much, compare to Mailbox. Without Mailbox around, I will have to manually setup my filters in a bunch of mailboxes.
The different between Spark submission tool and Hadoop is that, by default Hadoop uses all available machines (cores) for a single job, provide that input is big enough to be distributed into all machines. Spark, on the other hand, only use 2 cores for executors and 1 core for driver. In order to run Spark with all available cores, you first need to figure out how many executors there are on your cluster. I wonder what leads to the decision in Spark submission tool design.
It is very hard for me to think about what I’ve leaned by the end of a week. I’d better write and update a draft right after I learned something, when the memory still fresh. Oh yes, I still need to write two meeting summaries for this week, so I’d better get going. Hence the short blog post.