Quote:
Originally Posted by kiwidude
@drMerry - 18,000? Sigh. You wouldn't try to convert 18,000 books at once - Extract ISBN is doing a lot of the same steps underneath so I am not surprised it died. I cannot update ISBNs from the background as the database updates are single threaded, that is why it works the same way as metadata downloads in 0.8. Trying to "resume" a job after a Calibre crash is way more work than I can be bothered with to be honest, I would rather just not have it crash in the first place.
Did you see memory usage climbing before it crashed? The most likely explanation is a memory leak somewhere.
|
I love to look at (over) the edge of possibilities.
And of-course you should want to convert / run it in small parts, but if I had, this leak would not be noticed at this point
.
I did not check the mem, but I see you did it.
A (rather easy?) way to 'catch' this error (partly) would be to do the following:
Create a silent-run function. This function would run a scan and silently apply changes.
You than could use this function to create your own batch, and have a list of files todo whom you run in batches of 100, 1000, user-selected number of files at once. After each batch, apply changes, remove files from todo.