Previously, we had been storing partially-processed parsing entries in a file-system based "index". This set of patches replaces that logic with an analogous, database-based version. It continues to use os path separators and continues to be prefixed with an irrelevant directory path, but I've reached a good stopping point. Everything should be working without any regressions. Stepping through individual changesets is probably the best approach to reviewing.
Builds on #296(diff)
Previously, we had been storing partially-processed parsing entries in a file-system based "index". This set of patches replaces that logic with an analogous, database-based version. It continues to use os path separators and continues to be prefixed with an irrelevant directory path, but I've reached a good stopping point. Everything should be working without any regressions. Stepping through individual changesets is probably the best approach to reviewing.