cancelling massive konqueror file operations does not cause konqueror to properly cease disk operations

[N7DR]

Binary package hint: konqueror

This is in gutsy, but was in dapper (and presumably every release in between).

1. Create a large file hierarchy (by "large", I mean one that contains hundreds of thousands of directories and a million or more files).

2. Using konqueror, initiate a copy of the entire hierarchy to another destination.

3. After a while, hit cancel. (konqueror takes forever even to figure out what to copy, but that's a different bug)

4. konqueror will *appear* to have ceased the operation after you hit "cancel". At this point you can even quit konqueror and it will *appear* to close.

5. However, konqueror will continue to perform disk operations for many hours. Running "top" confirms that the process consuming the CPU time is konqueror; and the hard drive stays continuously lit.

This is completely reproducible.

(Running amd64 gutsy on a dual-core machine.)

[Harald Sitter (apachelogger)]

Just tried in KDE 4.1.2 with 90,000 directories and 360,000 files.
I once canceled while dolphin was trying to aggregate the nodes which left dolphin being busy for a couple of minutes, however it was indicating this properly, and as said, it was busy for a couple of minutes, not hours.
The other time I canceled it while it was actually copying, the processed quit immediately and dolphin was not using CPU anymore, thus was able to quit properly.

[N7DR (doc-evans)]

This bug was filed against konqueror. Does konq share the same code with dolphin, so that we know that the bug is fixed in konq as well?

[Harald Sitter (apachelogger)]

Yes