[BusyBox] undefined reference to `read_bz2' `BZ2_bzReadOpen'

[Glenn McGrath]

On Wed, 5 Nov 2003 14:36:35 -0600
Rob Landley <rob at landley.net> wrote:

> We know.  The bzip library was rewritten from scratch, and tar hasn't
> been updated yet.  It's on both my and glenn's to-do lists for
> whichever of us has spare time first.  (I had exams this past week...)

tar -j should compile and work now, i reverted back to the fork method
(decompress in child process) last time i was using the fork method i
recieved occasional bug reports about a "Short read", im hoping we dont
see that error this time, last time we were mixing streams and low-level
file descriptors.

Keep an eye out for "Short read" error's, when using tar -j, and a let
me know if you see any.

If i dont see any problems i will probably move gunzip back to the fork
method as well, uncompress(int in, int out) seems to be a more natural
interface for decompression code.

> Don't ask me why half of tar is in the libunarchive directory.  I
> don't know.  (If this is something other than tar, I'd like to know
> about it...)

There is lots of common code between ar, cpio, dpkg, dpkg-deb, tar, rpm,
rpm2cpio. They all extract data from an archive, the biggest difference
is the format the data is stored in.

These unarchiving applets define there functionality by assigning
functions to the various function pointers in the archive_handle_s
struct, allowing the actualy extraction to be handled generically
between all these applets.

get_header_tar* is used by dpkg and dpkg-deb as well.

I may try and refine the common unarchiving code post 1.0, but untill
then its just bugfixes.

dpkg and d-kg-deb are still broken, i havent fixed them yet.


Glenn