|
Yes - during the first stage all critical infomation from the calibre database is loaded into memory structures. Having said that, 500MB sounds extreme - on my own library with around 13.5K books all in several formats only about 120MB is used. Changing this approach is not practical as it would involve a complete rewrite from scratch. Having said that it is always worth revisiting the code to see if reductions can be made.
As to using custom fields this on the 'todo' list, but so far nothing satisfactory appears to be close to reaching completion. What is most likely to materialize is the ability to have a tag like named field treated in the same way as is currently supported for tags.
|