A program called CutOff

RodeOrm · Post by **RodeOrm** » 2003-01-05 08:38

Hello everyone

I have a program called: CutOff (It's only about 450KB) and with this program you can cut off the end of a file. This way you don't have to get any "Rollback inconsistency..." anymore. Someone called Zargblatt made it and I have no idea where I found it.

But if you're interested, e-mail me at: [email protected] and I will send it you. Maybe someone could put this up on his website ? It's a great tool for easy completing a download.

ivulfusbar · Post by **ivulfusbar** » 2003-01-05 08:42

is the program smart or dumb? does it handle mp3s different from mpeg etc? looking for frames and other stuff? or does it only cut off X bytes?

Sedulus · Post by **Sedulus** » 2003-01-05 10:04

ivulfusbar wrote:is the program smart or dumb? does it handle mp3s different from mpeg etc? looking for frames and other stuff? or does it only cut off X bytes?

dumb, it cuts off bytes, with simple syscalling, extremely fast

Sedulus · Post by **Sedulus** » 2003-01-05 10:15

RodeOrm wrote:Maybe someone could put this up on his website ? It's a great tool for easy completing a download.

I've put it on my website, and it will stay there unless Zargblatt mails me and tells me he doesn't want it there, or if this is so popular that my bw suffers.
(read the note about IE!)

/sed

TX · Post by TX » 2003-01-09 19:17

http://dc.ww-ei.com/files/Misc.%20Other/CutOff.exe <--- There you have it..

"Nice prog"

image · Post by **image** » 2003-01-14 01:11

if anyone can give me a source, i'll try to patch dc++ so it will cut X bytes atomatically if it encounters a rollback inconsistency.

Sedulus · Post by **Sedulus** » 2003-01-14 13:50

image wrote:if anyone can give me a source, i'll try to patch dc++ so it will cut X bytes atomatically if it encounters a rollback inconsistency.

I'm not used to win programming, so there are probably more MFC'ish ways to do this, but this should work:

Code: Select all

#include <io.h>
#include <fcntl.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <stdio.h>

   // char const * fileName = "somefile";
   // long removeBytes = 4 * 1024 * 1024;

   int fd;
   if( (fd = _open(fileName, _O_RDWR | _O_BINARY, _S_IWRITE )) < 0 )
      somethingWentWrong();
   if( _chsize( fd, _filelength( fd ) - removeBytes ) != 0 )
      somethingWentWrong();
   _close( fd );

#psyickphuk · Post by **#psyickphuk** » 2003-01-29 11:14

You can choose to rollback x bytes in DC++ anyway, is this prog neccessary or am I missing something?
cheers...

Post by **GargoyleMT** » 2003-02-02 22:45

#psyickphuk wrote:You can choose to rollback x bytes in DC++ anyway, is this prog neccessary or am I missing something?
cheers...

DC++ just re-gets the 4096 (or whatever your rollback is set at) bytes and compares them. It doesn't search backwards if there's a mismatch, it just determines there's an error and aborts the download (without having written any of the mismatched bytes to the file).

OmniXBro · Post by **OmniXBro** » 2003-03-06 08:23

well, this rollback inconsistency problem seems obvious and important enough, and the solution doesn't seem to be very difficult either (for you programmers ^^) so will you guys include this in the next update? (delete a few times the buffer size until the inconsistent bytes are all removed) but eh.... if it IS really not the same file, we don't want the downloaded file to be reduced to nothing by that, do we? ^^
lol, that would be funny.
Hey GargoyleMT, nice to see you too ^^ I never had the chance to speak to you about the rollback inconsistency problem, did I? ^^
annnnyway, I hope you guys fix it real soon

gambate!

Post by **GargoyleMT** » 2003-03-06 14:45

OmniXBro wrote:well, this rollback inconsistency problem seems obvious and important enough, and the solution doesn't seem to be very difficult either (for you programmers ^^) so will you guys include this in the next update? (delete a few times the buffer size until the inconsistent bytes are all removed) but eh.... if it IS really not the same file, we don't want the downloaded file to be reduced to nothing by that, do we? ^^
lol, that would be funny.
Hey GargoyleMT, nice to see you too ^^ I never had the chance to speak to you about the rollback inconsistency problem, did I? ^^
annnnyway, I hope you guys fix it real soon
gambate!

Well, this does pose somewhat of a problem, Omni... In order to roll back more, the transfer needs to be restarted... And there's no command for aborting a transfer in progress. So you need to disconnect and reconnect. Each time you rollback, you'll have to do this dance again. Maybe I'm just putting an ugly face on it, but it seems like there -is- no elegant way to do this currently. How many times should this happen? As someone mentioned before, they had to cut off the last 20m of a file. At 4k each slice, that's a lot of wasteful connections for both you and the person uploading to you. If more clients support upload queues, reconnecting instantly won't be possible - it's really hit or miss as is.

Patching the rollback buffer to work the way some of you need it to is a waste of time, compared to implementing a permanent solution: file hashes, specifically Tiger Tree Hashes. These would allow DC++ to verify chunks of the file as it gets them.

yilard · Post by **yilard** » 2003-03-06 15:05

GargoyleMT wrote: Patching the rollback buffer to work the way some of you need it to is a waste of time, compared to implementing a permanent solution

Right.

Everyone with problems:
Honestly, this mismatch happend 2-3 times to me during half year. (and I have downloaded hundreds of gigabytes). Maybe this has to do with which operating system you use. Is it Windows 9X or Windows NT/2000/XP?

nesteaboy · Post by **nesteaboy** » 2003-04-05 15:33

I think I am having a SEVERE case of the problem mentioned above.
I have a dedicated T-1 connection so I am able to download 10s of Gigs a day. It maybe that there's very high packet loss on my T-1 but I am getting TONs of files that end up "stuck" in my Incomplete Folder.

To give you a sense of what TONS mean, there were 4 Gigs+ and 45 Files I had to delete from my Incomplete Folder. I had to delete them because they were all had them same problem :

If the file is 15,536K, DC++ will download the file until the very last byte, then I would painfully watch the download speed go from say 70Kbps to 0Kbps and then eventually DC++ will remove the user I'm downloading from from my queue. The downloaded file, which would now be 15,536K minus 1 Kbyte would remain in the Incomplete Files FOREVER until I delete it.

I think this problem and the Rollback Inconsistency problem may be related (or the same). Is anyone having similar problems with similar magnitudes? I mean, 4 Gigs a night is a LOT of "upload" bandwidth to waste for some people. Is this something I can fix by setting the Rollback Buffer and the Write Buffer to a magic number? What would that be?

Thanks in advance.

yilard · Post by **yilard** » 2003-04-05 19:13

nesteaboy:

PLEASE, which dc++ version are you using? Which operating system?? If you want us to attempt help with rarely occuring problem, please include complete information....

Marvin · Post by **Marvin** » 2003-04-06 05:03

Well, it really looks like rollback inconsistency, and to be sure you'll just have to look in the Download Queue, the last column is called Errors.

The rollback inconsistency error in not related to resume, but to data loss : it can occur even if you download a file without interruption. Due to data loss, your version of the file is shorter than the source, so last bits don't match, DC++ detects it and displays the error.

AFAIK, rollback function will cure in only one case : the transfer has been interrupted and the error is within the rollback size. Raising the rollback value would make this lucky event more frequent, but it is not a magical cure (errors will continue to hapen before the first bit of rollback : in a 1TB file, the error could be on the 2nd bit).

Since we do not know anything about what hardware/software/OS you use, I will give you vague answers :

Had you a config that worked before, try to get back to it.
If you use Zone Alarm uninstall it.
If you use an old version of DC++, upgrade.
If you use the last version of DC++, try to get back to the preceding, then the antepenultimate, and so on...
Try to remove anything between DC++ and your ISP that could mess with packets.

Then give us feedback.

nesteaboy · Post by **nesteaboy** » 2003-04-07 03:28

Thanks for the help. Sorry for the lack of detail.

DC++ Version : 0.18 -> 0.241 (tried them all)

OS : Windows 2000 Advanced Server (fully patched)

Network Card : 3Com 100BT (it works, I've changed a few)
Hub : Intel Hub (it works, I've also changed a few)
T-1 Connection : Verizon

ZoneAlarm : YES!!! (I WILL REMOVE RIGHT AWAY). Does DC++ work with ANY firewall programs?

nesteaboy · Post by **nesteaboy** » 2003-04-07 03:36

And YES. Tons of Rollback Inconsistency Errors under Queue.
I've gone ahead and uninstalled ZoneAlarm. Will let you know if it helps!

nesteaboy · Post by **nesteaboy** » 2003-04-08 02:53

After uninstalling ZoneAlarm, it works PERFECTLY! Thanks a lot!
ZoneAlarm must have blocked a byte or two from time to time causing the total file size to be less than expected

MrBrain · Post by **MrBrain** » 2003-09-01 00:54

nesteaboy wrote:After uninstalling ZoneAlarm, it works PERFECTLY! Thanks a lot!
ZoneAlarm must have blocked a byte or two from time to time causing the total file size to be less than expected

And now you sit without firewall instead, not really a sollution that, eh?

Sedulus · Post by **Sedulus** » 2003-09-01 05:16

yes, and you'll starve to death when the McDonals chain goes out of business...

Twink · Post by **Twink** » 2003-09-01 06:18

MrBrain wrote:
nesteaboy wrote:After uninstalling ZoneAlarm, it works PERFECTLY! Thanks a lot!
ZoneAlarm must have blocked a byte or two from time to time causing the total file size to be less than expected
And now you sit without firewall instead, not really a sollution that, eh?

well he could now install a firewall that doesn't have a history of corrupting downloads on various programs.

hhacks · Post by **hhacks** » 2003-10-18 09:22

hejho!

do everyone who wants this proggie
current adress to get cutoff, at zargblattÃ‚Â´s hp
>> http://www.stud.ntnu.no/~dukefoss/CutOff/cutoff.html <<

ysy hhacks

DC++

A program called CutOff

A program called CutOff

Re: A program called CutOff

CutOff.exe

Re: CutOff.exe