International characters in name of .torrent & inside of it

[val_dam]

Hi.

I use 1.3.0_rc2 (deluged + deluge-web) on freebsd-8.0r-amd64

I've noticed:
1) deluge/deluge-web denied load .torrents named with non-ascii characters
(russian example: http://opensharing.org/download/51135/ );
after renaming it is quite ok with autoadd-directory feature

2) this .torrent contents directory with non-ascii characters (russian) name;
however deluged allocated and downloaded it correctly;
the problem is:
deluged doesn't move this .torrent from autoadd directory and can't add following .torrents from autoadd directory

[johnnyg]

Could you run deluged with debug logging enabled (http://dev.deluge-torrent.org/wiki/Faq# ... ingenabled) and paste any errors here.

[val_dam]

I couldn't repeat situation I described above,

but I got some other errors:
usually it happens at the moment of deleting (torrent+data) not fully downloaded torrents

1.3.0-rc2/FreeBSD-8.0R-amd64
deluged and delugeweb only

deluged.log and delugeweb.log attached

[val_dam]

Hi again!
I've eventually managed to repeat this situation many times and it isn't about international characters, it's about time, I suppose.

I have my "autoscan" directory shared by samba and I usually load .torrent-files with samba.

So, sometimes all is ok, but sometimes strange thing happens:
1) I load .torrent-file into "autoscan" directory with samba, it is accepted by deluge for downloading
2) .torrent-file isn't deleted(moved) from "autoscan" directory
3) deluged.log hasn't any records about this torrent
4) all followed torrents in "autoscan" directory aren't accepted for downloading by deluge, helps restart of deluge only

I think deluge sticks about first .torrent-file, because, perhaps, samba has relative long time for writing, and deluge has some glitches about processing files opened by others processes (like samba)


I had got situations like this in my own software, and solution was easy:
1) try to load(read) file, if error then write log-record and just skip it (go to the next file)
2) try to check data inside file, if error then write log-record and just skip it (go to the next file)
3) try to move/delete file, if error then just write log-record and just skip it (go to the next file)
4) accept data from file for processing (downloading)

if the problem is slow network operations, such file will be processed by one of the next iterations

[maxx.h]

the problem with unicode characters in the filename should be fixed in the GitRepo (will be included in 1.3.0).


The other problem still may be something that needs to be looked at.