Tyler Neely

Transactions are now in 0.26! They are implemented in an incredibly naive and low-performance way, and eventually may rely on this MVCC pagecache to become fully-featured. Keeping this issue open,...

Hey @Firstyear, all of those possibilities around tree-level operations become possible with transactional concerns being pushed into the pagecache! Applications can already rely on sled for their concurrency management layer,...

@Firstyear for the code you posted to work, with open_tree being transactional, everything the pagecache does needs to change from a lock-free approach to a batched transaction approach, and every...

@tomtau While this would be simple right now, if trees are parameterized by deserialized value types, it would no longer work. I'm intentionally keeping it totally type agnostic via tuples...

@ordian that should be pretty easy, as long as all trees are the same type.

@divergentdave I agree, that seems like one of the best places to start. The first module I started playing with is the serialization module, which has turned out to be...

Hey @jon-chuang, thank you for opening the issue! Indeed, I'm currently working on pointer swizzling, which has caused a lot of architectural changes that I've needed to step back from...

It's preferable to use our version compared to the actual ALICE tool because the actual tool is pretty heavy-weight, and by instrumenting our code directly, the overall complexity can be...

On discord, piotrek mentioned that in their research performing write-reordering tests targeting persistent memory systems, the most value was achieved simply by playing writes one after another, and in reverse...

After puzzling over the ALICE output, I believe it may be pointing out a vulnerability that has been floating around in the back of my mind: we only use a...