fix: ensure nested mount points are handled in the correct order

[icewind1991]

Otherwise etag propagation across nested mountpoints won't behave correctly.

To test:

• Create groupfolders gf1, gf1/gf2 and gf1/gf2/gf3
• Apply the patch, but change the order of $a and $b around to change the sort order to be intentionally wrong
• do a propfind with Depth: 1 on gf1 and on gf1/gf2
  • Note that the etag for gf1/gf2 differs between the two requests
• Restore the correct sort order by changing $a and $b back
• do a propfind with Depth: 1 on gf1 and on gf1/gf2
  • The etag for gf1/gf2 should now be the same between the two requests

[artonge]

Make sense.
Any idea how they could be in the wrong order in the first place?

[icewind1991]

Any idea how they could be in the wrong order in the first place?

Did some more testing and it seems to be just the order the groupfolders were created in.

So if you create gf1/gf2 before gf1 it would trigger the bug.

[icewind1991]

/backport to stable28

[icewind1991]

/backport to stable27

[icewind1991]

/backport to stable26