* Fixes a couple of memory leaks (although not dangerous in practice, since we are talking about objects with a lifetime up to the end of the application)
* Fixes heap use after free
PR #19650.
Closes#19632.
Improves "Metadata received" stop condition handling by "Add new torrent" dialog.
The problem is when "Metadata received" stop condition is set the metadata can still be received 0.1 sec before clicking the "OK" button so torrent is actually added with metadata and therefore it cannot respect "Metadata received" stop condition.
The solution is to uncheck "Start torrent" checkbox once metadata is received.
PR #19597.
Closes#19583.
PR #19496.
* Add torrent entry status to represent tracker error
* Add torrent entry status to represent unreachable endpoint
* Display tracker entry next/min announce time
* Reset tracker entries when torrent is stopped
Warning: users are at their own discretion to create .torrent with >= 64 MiB piece size as not
every torrent client supports it.
Larger piece sizes are only available when using libtorrent 2.x. libtorrent 1.x is not
efficient with memory usage and in order to avoid user complaints it is limited to 128 MiB.
Also note that, as of this writing, libtorrent 2.0.9 has an internal limitation that only
allows loading maximum 256 MiB piece size. And therefore > 256 MiB size options are forbidden
for now.
Closes#19527.
PR #19535.
The change is in preparation for adding the possibility to create
torrent files via the API.
Rework TorrentCreatorThread to be a more lightweight QRunnable class.
The parameters are now defined on construction time and are fixed
throughout the lifecycle of the TorrentCreator. The lifecycle of the
object is not bound to the one of QDialog anymore; it is now handled
by the QThreadPool. This will enable easier queueing of multiple torrent
creation jobs without risk of spawning many threads.
PR #19500.
PR #19468.
* Use QHash to map tracker endpoints
* Don't clear numPeers unexpectedly
* Remove outdated tracker entry endpoints
* Move presentation logic from Core to GUI code
* Show all endpoints per tracker in tree structure
---------
Co-authored-by: Kacper Michajłow <kasper93@gmail.com>
Changing the data may affect the layout of the sort/filter model, which in turn may invalidate the indexes previously obtained from selection model before we process them all. Therefore, we must map all the selected indexes to source before start processing them.
PR #19372.
Closes#19359.
