...
removed tab Database
Changes in Behavior
paralelisierung
Schma file caching. Not reading every time, reaction on file changes event
TODO multiline index file handling, split every times
SID mapping
...
Performance improvement and multi lile index file handling
The main objective for this major change was performance optimization. Until DocumentRouter3, index file processing is single-threaded.
This means that the index files are treated in parallel, now. In order to fullfil stiff the max-fetch-file-feature we needed to redesign multiline index file handling.
In version 4, if an multi line index file is found, it will be first splited into single line index files. Therefore the error handling for such files has changed. THe multiline index file is splitted and after successful split it will be moved to Success folder (or deleted). If it fails, created index files are gone and the multi line index file is moved to Failure (or deleted). After the split, the single line index files are handled normal.
Another change is about the schema file reading. In version 3 on every index file read the schema was read and parsed again. In Version 4 the schema file reading and parsing is done once at the start of the instance. The parsing result is cached. This cache is deleted if schema file change takes place. This mean as well, that if the schema file is changed for another instance, all instances will reread its schema. So we could maintain the old behavior that the schema file change is effective on every file handling.
SID mapping
TODO