On Wed, 24 Jun 2009 16:51:09 -0300, you wrote: > >But it also has this code in file_xfer.cpp (error checking removed to make >email shorter) > >// for downloads: if we requested a partial transfer, >// and the HTTP response is 200, >// and the file is larger than it should be, >// the server or proxy must have sent us the entire file >// (i.e. it doesn't understand Range: requests). >// In this case, trim off the initial part of the file >
Wouldn't this mean the client already handles this correctly, as long as WCG re-configures their server to ignore the range-request and sends the whole file? But, it won't fix the problem that WCG broke the download-resuming. While download-resuming doesn't matter much for anyone sitting with a 1 Mbit or faster connection, anyone that's still stuck with dialup on the other hand won't like WCG is forcing them to re-download any broken download. Also, would guess the odds of having a connection-error breaking a download is largest just for the dialup-users, the users that most needs the download-resuming... So, any changes that breaks download-resuming is in my opinion a bad change, and should be avoided. > >(Which seems quite silly in my opinion. Why not seek to the beginning before >saving any data, instead of appending and then "trimming beginning of file" >after the transfer is done? And why is a temporary file needed to trim the >beginning?) > Just a guess, but ocassionally on various web-pages, then downloading the download-progress shows "unknown file-size"... If the server doesn't tell you the file-size and so on, but just starts serving you the file, how do the BOINC-client know it started at the beginning, and should delete everything? To me, the code-snipped just seems to also take into account servers that isn't telling this information. -- "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
