actually I have had very few time to test your recent updates to wimlib, and to improve ImagePyX.
However, I've been thinking about a possible improvement: one useful tool to implement could be a "binary repairer", an app able to scan a broken WIM and detect/guess position of valid XML data and file resources, and, finally, reconstruct a valid WIM or some of its images.
The 1.4.2 release is just a couple bugfixes I had pending, so there shouldn't be much to test compared to the 1.4.1 release (which had a lot more changes). I've been busy with other things for the most part.
A binary repairer sounds interesting, although I haven't had this feature requested before. I'm not sure how often it would actually be useful to someone. In the event of file corruption due to a failing hard drive or otherwise generally you would want to restore from backup (if one exists, of course...) rather than use the corrupted data.
At some point, those guys at Info Zip put the "zip -FF" sub-command to recover broken ZIP archives: very few ZIP archivers can do that.
Surely, something similar would distinguish well wimlib-imagex from its MS counterpart, and give it a superior pedigree...
Another useful thing could be an option to pre-select more quickly the files to append to a new image, by comparing their date× only instead of taking their SHA, in a way closer to common file archivers (this should speed-up common backup operations, preventing the analysis of the full input before the append pass).
Hi, just to follow up on this: The last feature you suggested (pre-selecting files to append) has been implemented in wimlib v1.5.0. In wimlib-imagex this feature is exposed through the --update-of=[WIMFILE:]IMAGE option to wimlib-imagex capture and wimlib-imagex append.
It is also possible to create a "delta WIM" that contains just the streams not already present in some other WIM file.
Repairing WIM files has not yet been implemented.
Nice feature, thanks! This will really improve documents' backup speed in some scenarios of interest of mine.