There's free space check in CopyJob. And Dolphin still gives an error in the said situation. But if some application decides to directly use FileCopyJob then IMO the free space check should be at application's side. Anyway, I tried adding it here and it felt awkward to me.

Opened Dolphin, Ctrl+C on a file, Ctrl+V it into the same folder, let it sit at the "Override?" prompt until the job view in Plasma showed up. >Chose a different filename and continued.
[...]
The URL isn't updated in case of a rename but I bet it just needs a m_bURLDirty call somewhere in the conflict handler

Remove unrelated changes commited by mistake.

Delete .part file when Overwrite flag is set.

Print strerror.

Restore sourceStated in CopyJob.

@dfaure ping

Update.

Skip unit test on windows.

Removed warning message.
Removed arg().
Added test cases.
Used QFile::resize().

Fixed all issues.
Added test case for suspend.

@cfoster Thanks for the patch :)

@ngraham The destination file is corrupted as soon as the user clicks overwrite. This patch deletes that file. Is that file supposed to be preserved (to prevent user panic)? My suggestion is to append ".part" or some similar extension when an overwrite starts, so the user knows that there data is gone.

LGTM

Add bug id.
Remove old code.

In T8075#174716, @ngraham wrote:

What needs to be done to push this forward? Do you need any help?

@bruns ping

@kpiwowarski Still WIP :(

@aacid Add an action to "Tools" menu for accessing the signature panel. Optionally add an action to verify all signatures but I feel it's redundant due of the message widget.

@aacid Sorry for late reply. This got lost in my inbox.

They emit a generic message. A descriptive message can only generated from within job.

Scraped everything but the changes from file_unix.cpp

In D17737#381029, @bruns wrote:

I think it would be much simpler if you just tried to do the the FICLONE iotctl in the job, without any prior checking:

• no possibility for races
• less syscals
• less code

update

Removed vector initialization.
Increased vector capacity in constructor.
Appended {0,0} at first.
Stored time and size before appending to prevent extraction.

Used a structure to group size and time.

Used QVector

In D17528#376349, @dfaure wrote:

I'm surprised, how can a QLinkedList (with nodes allocated all over the memory) be better than a static array (which fits into the same memory cache) ?