|
I think you may need to do an lsof during "Processing" to see if you can find the other end of the pipe, if busybox supports that. Or can you look through /proc if you're really quick?
Using the widely available unzip library code, a caller can open a zip/epub file, and then step through the zip file's internal structure in order to access any given file. This could be more efficient if the OPF is at the end of the epub (zip files are walked from the end of file backwards!), but even if it is not it is a few seek()s and read()s away at most. You absolutely don't have to unpack the entire file.
So your testing with find and unzip really is worst case. Yet still faster than Kobo!!
Have you tried updating an sqlite database? I think you said earlier in this thread that sqlite perf was pretty bad on the Kobo.
|