behead/truncate large file

Hi,

I have a pretty big wav file and want cut off a small bit of the front, or a small bit of the back. To get and idea of size says its 2MB front or back bits from a 1GB file.

So there are tools available, such as sox for this. In fact there's a number of ways .. but most involved creating a new large file (less the bits taken off): transfering to disk and overwriting the original file with the new truncated file. Which is all very slow.

Seeing as the majority of the data is not changing at all, I was wondering if it wasn't possible to leave the majority of the file as is, and recreate the boundaries (front or back) of the file in some way, to make it all much quicker.

I was lookign for suggestions. Right now I'm looking at the possibillities of mmap, and then I'll look at any apis that the ext2 filesystem may offer. It's not so important about it being a wav, even doing it to a text file is fairly interesting.

Many thanks in advance !

Is it doable, I would guess yes.
Is it wise to do, I would guess no.

If you make, a mistake you could destroy all the data on the hard drive.

Because, the only way to not have to re-write the file will be editing the file structure info on the hard drive.
I would start by learning as much about the ext2 file system; maybe it has a simple and safe way to do it.

In NTFS, I believe it would be too dangerous for the slight gain in speed.
But, I only know how the MSDOS file structure was done. Never looked at the logical disk geometry for NTFS or ext2.

Tim S.

Even file system manipulation would only help for the trim the beginning case where the new beginning happened at an allocation unit boundary. Files cannot begin at arbitrary positions on the disk (assuming, of course, you don't want to write your own file system to handle this...)

Many thanks stahta01 and Cat for your suggestions. I agree with both of you this "file boundary modification" seems a mad idea now.

Equally, I also think that - in general terms - is there is a growing trend towards bigger and bigger files. I got to thinking that perhaps the way is not to modify the file at all, but merely to access it differently, using offsets and such, fseek() and friends. A good example could be Youtube, some of the files now going up are huge, so to draw people attention to a certain part, you can submit a comment prefixed with the time point of the section and this will render as a link when the comment is published.