Here is my testing of this patch. I picked 10 sets of 50 random packages from the Arch Linux repos: pacman -Sql | shuf -n50 Then I cleared all caches, and ran: pacman -Sddw --noconfirm I ran each list 5 times, alternating between patched and unpatched versions of pacman. The medians of these five runs are: Unpatched With Patch 17 21 12 10 33 33 29 26 34 29 47 36 30 28 31 25 32 29 50 40 Expressing these as a percentage of the unpatched time (so less than 100 is an improvement): 124 83 100 90 85 77 93 81 91 80 For an average of 90% - meaning on average the downloads are 10% faster when we order the packages. This is obviously based on my connection, my mirror, and my choice of parallel downloads (5), and with limited numbers of trials. But as it stands, I think accepting this patch will improve total download speed. I will properly review the patch soon, with the aim to accept unless contrasting results are provided. Allan