ntdls / ntdls.katzebase.server Goto Github PK

Fire up the thread pool and add a hash set that contains the queue/completed count for each GUID based workload. This was we can keep the thread pool alive and even share it across queries. The pool can also be safely grown / shrank depending on load.

Allow it to add fields that do it currently exist.

In UI, the run state and results tab should be part of the tab page. This is so multiple scripts can be run at a time.

When running script from UI, trim all lines, ensure trailing CrLf and split on "CrLf".

This can assist in tracking down json files.

Determine if we enable IO deferment on system created transactions. I think we can.

Implement stored procedures that take parameters.

The “management” classes for schemas, indexing, documents, etc are very messy. Separate the methods by responsibility.

Allow to query for running processes and view blocks.

Add .net functions (like procedures) that take parameters and are stored as a compiled assembly.

When collapsing a condition formula, check to see if either side is a constant or a field and treat it accordingly. This way the constant value can be on the left or the right. It also means we could then use constants in a join condition.

Add options for flushing IO writes to disk. This can be enabled for LOG or DOCUMENTS exclusive of each other. Use this: . Enable file system write-through caching: The write-through caching policy ensures that every write operation is immediately committed to the disk. You can enable this policy using the FileOptions.WriteThrough flag when creating the FileStream object.

Also look at FileStream.Flush(true).

Implement built in scaler functions: GetDateTime(format), GetDateTimeUtc(format), Random(min, max), Guid(), LeftString(n), RightString(n), SubString(start, length), LengthString(string), PadString(character, length), Sha256String(string).

Add global min threads, max threads and thread multiplier configurations.

Create winforms application to export SQL server to Katzebase.

this removes a field from all matched documents.

Idea: Order the lookups by whether they have joins, then by the descending number of expected records.