Closed fpom closed 5 years ago
Hi! Any chance you’re using GStreamer? If so, we may have just fixed the problem in audioread: https://github.com/beetbox/audioread/pull/84
I have no idea whether I'm using GStreamer or not. How can I check? I just do: import acoustid
then acoustid.fingerprint_file(somepath)
.
BTW, I've replaced acoustid.fingerprint_file()
with subprocess.check_output(["fpcalc", ...])
and it's much faster, which surprised me.
It's actually pretty hard to tell—you can check whether you have GStreamer and the appropriate Python bindings installed, and then it's likely that's what's happening.
I'm not sure but I think is is installed: the following does not fail
import gsi
gsi.require_version('Gst', '1.0')
from gi.repository import Gst
So, since beetbox/audioread#84 has been fixed, my understanding is that the memory leak I've observed will be fixed in the next release?
Yep!
Nice ! :) Thanks for your help.
I'm using acoustid.fingerprint_file() repeatedly on a collection of MP3 files, and I see the memory usage of my script growing over and over despite the fact I store the obtained fingerprints in a DBM. (I've done the same with SHA512 hashes and I observed no such increasing memory usage).
So I strongly suspect a memory leak in the library.
I've installed 1.1.5 with pip, using Python 3.6.2 under Linux.