strata | Evented I/O B-tree for Node.js | Runtime Evironment library

strata is a JavaScript library typically used in Server, Runtime Evironment, Nodejs applications. strata has no bugs, it has no vulnerabilities, it has a Permissive License and it has low support. You can install using 'npm i copacetic' or download it from GitHub, npm.

The b-tree package exports an object I like to name Strata. TODO Force the naming. In order to create a Strata b-tree you need to choose a storage strategy, you can store to either a write-ahead log or into a directory tree on the file system. Let's start with the file system. We'll be introducing different modules as needed in the final draft, but let's dump them all into the README.md for now. For our README.md examples we'll need to create some file paths. When you create a Strata b-tree you need to provide functions that will convert the objects you want to store to and from Buffers. Strata will writes those buffers to the file system or write-ahead log. You will need one serializer and deserializer pair for keys and one serializer and deserialiser pair for records. Records are inserted into Strata as an array of objects. This allows you to store an array that contains both JSON serializable (or otherwise serializalbe) objects and Buffers. We call this a parts array. WHen you serialize a record you will be given an array of parts and you must return an array of buffers. The length of the array of Buffers does not need to match the length of the array of parts. Similarly, to deserialize a record you provide a function that receives an array of buffers and returns an array of parts, that is an array of JavaScript objects that are maaningful to your application. Key serializers... write about this, please. When you create a Strata b-tree you need to provide two functions that will define how the tree indexed. The first function is an extractor. This function extracts a key from the stored record. Strata is not a key/value store, it is a record store. The key for a record in the store is extracted from the stored record. To extract the record you provide an extractor function. Strata works with compound keys. These compound keys are represented as arrays. All Strata keys are arrays. Your extractor must return an array that contains the values of the compound key. The array returned must always be the same length. WHen you insert data into a Strata b-tree you insert an array of JavaScript objects. The extractor function takes this array of objects and extracts the key values into an array that creates a compound key. Our initial example extractor merely returns a single element array, a compound key with one component. To complete the index we need to provide a comparator. A comparator should compare two arrays and return less than zero if the first array is less than the second, greater than zero if the first array is greater than the second and zero if they are equal. The arrays are equal if they are of equal length and the array element for each index in the array are equal. The first element in the first array that is less than or greater than the correspondding element in the second array makes the first array less than the second array. If one array is shorter than the other array and elements of the shorter array are equal to the correspondding elements in the longer array, then the shorter array is less than the longer array. With this sort function we can perform forward and reverse inclusive and exclusive searches against the compound keys our b-tree using whole or partial keys. To create a comparator I use Addendum which prvoides a comparator builder function that builds a comparator according to the aforementioned algorithm. In the above we have created a comparator that compares ... Strata stores data as an array of buffers. Deserialization converts that array of buffers into an array of object. Serialization converts that array into an array of objects. The extractor function accepts an array of parts. The parts are also user defined. To both insert into and retrieve objects from the tree, we must first search the tree to arrive at the appropriate page. To do this we use a Trampoline so that we do not have to surrender the process to an async call if all the pages are cached in memory. When call search with a Trampoline instance, a key and a callback function. The function is called with a Cursor object only. (This is not an error-first callback function from the good old days of Node.js.) The function is synchronous and all operations on the page must complete before the function returns. The synchronous callback function is a window in which you have sole control of the in-memory b-tree page. You should not hold onto the cursor and use it outside of the synchronous callback function. These operations are are verbose, but as noted, they are usually encapsulated in a module that provides the user with an abstraction layer. Retrieving from the Strata b-tree is similar. You invoke search with a trampoline, a key to search for, and callback function that accepts a cursor object. The synchronous function is the window in which you have sole control over the in-memory b-tree page. You should copy the values out of the in-memory page for use when the function returns.