On Sun, Oct 3, 2010 at 3:58 AM, todd rme <toddrme2...@gmail.com> wrote: > On Thu, Sep 30, 2010 at 7:30 AM, Matthias Fuchs <ma...@gmx.net> wrote: >> Hi, >> >> When moving and copying multiple files it can be quite tedious to make out if >> there are differences for all these files. >> >> >> >> ====Use Case:==== >> You copy hundreds of text files, knowing that most are the same, but not all. >> Now you are greated with multiple "Do you want to overwrite XY size Z with XY >> size W" dialogs. >> >> ====Proposal==== >> What I propose is to not show this dialogs if both files are identical, in >> the >> case of copying nothing should happen then, while in the case of moving the >> source file should be deleted. >> >> To check if a file is identical this should happen in a two step process: >> 1. Both file sizes equal and smaller a fixed size >> 2. Calculating the checksums for both files, the check for the fixed size >> above avoids long lasting calculations >> >> If 1. turns out to be false a dialog should be shown. >> >> >> This could be either opt-in (via a checkbox) or always on with just an >> information text in the dialog. The hash function should be one that is very >> fast to calculate and if the file system supports and stores checksums for >> files those should be used. >> >> ====Open Questions + Discussion==== >> What do you think of this idea, should something like that be implemented? >> Also what do you think of the Nepomuk Ressources associated with the files? >> Imagine both files have a different rating, what should happen then? >> >>>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe >>>> << >> > > I posted brainstorm forum idea about this last year: > > http://forum.kde.org/brainstorm.php#idea39563_page1 > > I proposed a three-stage process, similar to yours but with an extra > stage, an optional byte-by-byte check. > > Checksums are fast for small files, but they can take longer on large > files and on older systems. They also, as I understand it, are not > perfect. So I think that a better approach is that, for files under a > certain size, an automatic three-stage approach is used. First the > file size check, then checksum, then byte-by-byte. If all of those > pass, then the file is just deleted. > > For slightly bigger files, where the checksum is fast enough but the > byte-by-byte is not, only the first two stages are used. If they both > pass, the "File Already Exists" dialog box should be changes to tell > the user that the files are "probably" the same, and gives them the > additional option (on top of renaming, overwriting, and skipping) of > doing an "Exact check" (or something along those lines), which then > does the byte-by-byte check. If that passes, then the file is > deleted. > > If the file is really big, then even the checksum is not done > automatically. If the files have the same size, the user is told the > files have the same size, and the user has the additional options of > doing a "Quick check" and "Exact check" (checksum and byte-by-byte, > respectively). If the checksums match, you are back to to the > previous situation where the user is given the option to do the exact > check or do one of the standard actions. If the detailed check > passes, then the file is deleted. > > The issue with the nepomuk data is an issue even without this. When > you are moving files and decide to overwrite conflicting files, even > if they aren't the same. A simple check box for "merge nepomuk data" > or "merge tags" or something like that (if they both have data, of > course) would be very useful independent of this.
Sorry for dredging up such an old topic, but I was wondering if this might this make a good GSOC project. -Todd >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
