Open nekohayo opened 6 days ago
It happens on newer versions of aiohttp (iirc), but i haven't figured out why yet
Could it be related to one of these old issues? Otherwise I suppose your best bet is to ask in their discussions area if they might know of something that recently changed regarding this, I guess 🤔
I don't know if this is something you will want to consider, but reporting it separately just in case, and because I doubt that this is the cause of the performance problem seen in #29…
I have now isolated the conditions under which the tracebacks can be encountered. In my case, it only happens when running the non-flatpak version, such as the 1.3.0 package available in Fedora 39 (whereas @yochananmarqos encountered the issue on Arch Linux instead), even if that's the app version number as the flatpak version.
Beyond the Python tracebacks, the symptoms are that the upload speed will be reproducibly much lower than reality, even compared to LibreSpeed's website, because the upload speed measurement will "start" at the correct speed (~60-70 mbps) and then gradually go down towards zero (in my case, 10mbps) as the tracebacks occur:
…when running the flatpak version of the app instead, the upload speed will match other speedtests:
Behind the scenes, the non-flatpak version causes these tracebacks to occur during the upload speed test:
(repeated thousands of times)
Then:
Maybe the dependencies are packaged differently, or maybe you depend on some newer (or older!) Python library features that have not been reflected in the dependencies metadata?
Version numbers from the Fedora 39 package and its dependencies are: