[Dctc-general] Rollback resuming?

[ful hasch]

Hi
Sometimes when I have downloaded a large file from several sources it has 
been corrupted even if all the source files were ok.
Today I found a page ( http://www.geocities.com/divx_repair/ ) that might 
have a solution to the problem.
It says:
"When a filetransfer gets broken unexpectedly on the server side, your file 
may contain some bad bytes at the end because the client didn't detect the 
error immediatly. Here is an extract from the help file of the popular 
downloadmanager GetRight:
Rollback XX K on resumed connections: Because some data may have been 
corrupted when GetRight was disconnected, this allows you to backup a little 
bit and reget a small amount of data to be sure that no errors are in the 
file on your computer."
I don't know if this would be hard to implement into dctc or even if it 
would be a good thing. It should certainly be optional.
I've also got some observations about the behaviour of GDL autoscan. When I 
generate an autoscan pattern, the GDL starts downloading from several 
sources simultaneously even if I have set the max number of running sources 
per GDL to 1.
And if I generate autoscan patterns to several GDLs at once and the min 
delay between two searches is greater than 0, only one of the GDLs will 
receive search results. I can of course stop and detach a GDL and reattach 
it to find new sources, but that's not very practical if you have dozens of 
files with autoscan patterns.

/fulhasch

_________________________________________________________________
The new MSN 8: smart spam protection and 2 months FREE*  
http://join.msn.com/?page=features/junkmail