Repository navigation
Replies: 5 comments 8 replies
|
I have never used this tool, but I'd like to test it before studying a way to implement it. I already have an idea on how to implement this... but in first I'd like to know more about. I use Debian. Please provide me some examples. |
|
@dacampsss While I agree that the download speed could be improved, the real issue here is that kdenlive does not support AppImageUpdate With that method updates only take a franction and you don't download the entire appimage over and over, we actually have a fork of appimageupdatetool where you can see a simple benchmark of me updating CPU-X and using only 2.65 MIB |
|
@Samueru-sama Does AM already support appimageupdatetool flow for partial downloads? I see it in some of the install scripts, but it seems like it still does wget for downloads. I thought zsync was the tool recommended for partial downloads? I tested probono's zsync2 to update some appimage and it worked well. Only needed to download 20%-30% of the file (maybe even less for small updates). Saves huge amounts of bandwidth. @dacampsss Using aria2 may not be a good idea, as it will blast servers with multiple simultaneous partial download connections and may result in temporary IP bans for some users. Also, it still has to download the entire file. |
|
I see, the wget command will just skip the download if the file is of equal size to the remote file. Looks like appimageupdatetool is very similar to zsync2 in code, and does a pretty good job. Nice. |
There are two separate questions tangled together here — download speed and which tool — and they have different answers. I looked at how AM currently fetches things to ground this. Current state: AM already uses On speed (the original request): the win @dacampsss is after comes from aria2's multi-connection segmented downloading — splitting one file across several parallel connections. Neither The catch with aria2 is that it's a dependency almost nothing ships by default, which works against AM's minimal-dependency design. The portable way to offer it without forcing it on everyone would be optional detection, the same pattern the scripts already use for One honest caveat: a slow download from a specific server (Kdenlive's, here) is often server-side geo/throttling. aria2's parallel connections only help if that server honors range requests and doesn't throttle per-IP in aggregate — so even aria2 may not speed it up in this particular case. On curl as a tool choice (a different motivation): there is a reasonable case for curl, but it's about dependencies, not speed. Since curl is already required for the metadata lookup, moving the download line to Short version: curl for dropping the wget dependency = reasonable; curl for speed = no. The speed ask only moves with an optional aria2/axel fallback, and even then the gain is conditional on the server. Available for follow-up in the thread if there are questions. |

Uh oh!
There was an error while loading. Please reload this page.
For some reason, for example, to download Kdenlive from their servers right now is a real challenge here where I live. I think parallel downloading using aria2 would ofc increase this process.
Thanks in advance!
All reactions