FlushRegion [Previously Source Code Part 2]: "Oh, I may look like a class called FlushRegion, but I am actually just _holding_ this class in case it would be useful to work with. So, you can call me by its name, but unless we're turning into an AI or something I don't think the classes are talking in words right now." Programmer 2 [previously FlushRegion]: "Is FlushRegion your idea?"

Programmer 1 and Programmer 3 and Programmer 2 all hold hands with a bunch of other programmers. "We want to implement an append-only tree structure for random access data with revision history! That's the story we're writing right now!" They all say this in unison.

1: "Let's just sit and think about it a little, maybe get to know each other. I see your shirt."

4 ["Linked List"]: "There are a few of us here. 2 is "FlushRegion", 5 is "ChunkRange", 3 is "Tree of Append-Only Flushes". 3, why don't you change your shirt to say "Flush Index"? That's more similar to class names." 3 ["Flush Index"] (previously Tree of Append-Only Flushes): "Okay !" 1 : "I'll be Tree of Append-Only Flushes!"

1: "Okay, so FlushRegion is a class for edges of a linked list, roughly. Each Flush Index is a length of them. FlushRegion sits between nodes, which are either writes of data, or terminators." 4: "There's been some confusion as to whether FlushRegion should be made of ChunkRange's, or entire Flush Indexes, or what. The confusion has been complicated by ideas around refactoring the Flush Index and changing the name of Flush.Entry to ChunkRange. We propose filling FlushRegion with ChunkRanges, because it is simpler to consider, involving fewer new concepts, new names, and new connections between concepts."

Programmer 1: "I looked back over the minutes and surveillance tapes" Programmer 2 ["FlushRegion"]: "We have those?" Programmer 1 ["Considering Append-Only Trees"]: "Do we? Anyway, I think my shirt should say Considering Append-Only Trees. I like this label." Programmer 2 ["FlushRegion"], Programmer 3 ["Flush Index"], and Programmer 4 ["Linked List"] come together and say: "We like it too!" Programmer 5 ["ChunkRegion"]: "I want to do more than just Chunk Region. I like the idea of working on those old concepts _around_ the Chunk Regions. The ones that for some reason people find a little confusing now."

ChunkRange (Flush.Entry) Concepts: "There was an idea of making the flushes have similar properties to the "dat project"'s protocol, where the trees are built full, binary, and balanced, to keep indexes small. I had to go look through my own records to find this." Considering Append-Only Trees: "Right. If we consolidate the ChunkRanges between the writes, we can keep the indexes really small. This was the big idea."

Building Append-Only Trees: "Let's implement the ChunkRanges between the Writes. What is needed first?" ChunkRange Concepts: "I'd like to look at the code we have so far."

Building Append-Only Trees hoisted a giant class called FlushRegion, and carried it into another class, the Flush Index class. Building Append-Only Trees: "You can be called Flush.Region now!" A friend of theirs changed their shirt to say "Flush.Region" instead of "FlushRegion". But the person wearing the shirt saying Building Append-Only Trees was only just getting started!

Once upon a time, long long ago, in a sunny glade far far away, a tree for random-access data on append-only storage was building. It was a happy young data structure, with a handful of substructures for interconnecting its parts, and a few methods for mutating and inspecting it.