Posted July 06, 2018
Generally my fork should
Faster for updating the manifest, especially for full manifest updates (becaue it doesn't pause between every file for significant time) particularly if you have games with lots of individual files. This may be less true for update only / new only updates because the resume is much more paranoid because touching the games clear the flags but for games with lots of files / extras it should still be a net benefit.
A little slower to download the actual game files (due to preallocation and download resume support requiring MD5 checking of chunks and downloading the MD5 chunk file). But these should be a net benefit overall (defragmentation avoidance should be very useful on Windows non-SSD drives over time) and download resume should be a massive help on Linux which can have some very large file installers.
The first verification should be a little bit slower but later verifications should be much faster since they only verify things that have changed by default.
Significant deviations from this probably indicate a problem of some kind so I'd like to know details.
Faster for updating the manifest, especially for full manifest updates (becaue it doesn't pause between every file for significant time) particularly if you have games with lots of individual files. This may be less true for update only / new only updates because the resume is much more paranoid because touching the games clear the flags but for games with lots of files / extras it should still be a net benefit.
A little slower to download the actual game files (due to preallocation and download resume support requiring MD5 checking of chunks and downloading the MD5 chunk file). But these should be a net benefit overall (defragmentation avoidance should be very useful on Windows non-SSD drives over time) and download resume should be a massive help on Linux which can have some very large file installers.
The first verification should be a little bit slower but later verifications should be much faster since they only verify things that have changed by default.
Significant deviations from this probably indicate a problem of some kind so I'd like to know details.