Re: [Bug-tar] Re: GNU tar Feature Question

[Sergey Poznyakoff]

Aaron M. Renn <address@hidden> wrote:

> I haven't looked at the code, but I suspect that it probably gives up
> after a period of time.

Nope, it does not. It keeps reading new records until it finds one that
is a valid archive header. In your example it obvioulsy could not find
one.

> My previous patch [...] performed the highly simplistic
> strategy of just advancing one byte at a time looking for something it
> could extract - till the end of the file if necessary.

Yes, that's the same strategy cpio is using when trying to restore
a broken tar archive. It works fine, indeed. I plan to implement it
in tar when I am through with more urgent issues.

Regards,
Sergey