787 was fixing the issue #786 not just at the location of the problem, but this reached the File class as well, modifying it to raise an unblob specific error, effectively aborting processing in a "well known manner".
I think this was wrong, and File worked as expected and was a rather generic class, and should remain so.
The fix also had changes in gzip and 7z handlers, ignoring empty files from the collected files, but that change was neither supported by tests nor explained - I think that was a redundant fix for the problem.
The root of the issue is not in the File class, but rather that the raised ValueError is not handled by the handler.
This MR reverts some changes and fixes only the failing function, covered with a test.
787 was fixing the issue #786 not just at the location of the problem, but this reached the
Fileclass as well, modifying it to raise an unblob specific error, effectively aborting processing in a "well known manner".I think this was wrong, and
Fileworked as expected and was a rather generic class, and should remain so. The fix also had changes in gzip and 7z handlers, ignoring empty files from the collected files, but that change was neither supported by tests nor explained - I think that was a redundant fix for the problem.The root of the issue is not in the
Fileclass, but rather that the raisedValueErroris not handled by the handler.This MR reverts some changes and fixes only the failing function, covered with a test.