Re: Announcing xorriso-0.4.2
Joerg Schilling wrote:
> What you implement may be a nice toy
It is a backup format that can compete with
programs like afio or star when it comes to
fidelity of file properties.
Its multi session capabilities allow fast
and easy-to-use incremental backups.
It can restore what it wrote. Alternatively
one can mount the images as iso9660 and will
have access to file name tree, most POSIX file
properties, and all data file content.
This release is about xorriso's ability
to verify backups. E.g. it checks the blocks
with the directory descriptors for matching
MD5 before loading them.
> > * New option -md5, new -as mkisofs option --md5
> Mkisofs does not have a -md5 option.
Note the double dash with -as mkisofs --md5.
There are some other extra options which help to
beef up a mkisofs CLI client to backup grade quality.
--quoted_path_list FILE File with list of quoted pathnames to process
--hardlinks Record eventual hard link relations of files
--acl Record eventual ACLs of files
--xattr Record eventual user space xattr of files
--md5 Compute and record MD5 checksums of data files
--for_backup Use all options which improve backup fidelity
> In case that mkisofs will eventually add a -md5 option, I cannot grant that
> this option will be compatible to what you implement.....
Now that you are aware of my --md5 it is
your decision alone whether to implement
a collision or not.
> > "cdrskin does not contain any bytes copied from cdrecord's sources."
> This is not true. You even admit that
> is: .... borrowed HEAVILY from cdrdao.
cdrdao is not cdrecord.
Actually the code in lec.c was introduced
by the founders of libburn under GPL.
So they admit, not i.
lec.c is not in use with SAO or TAO mode.
Whether libburn's CD RAW modes are good for
anything has still to be determined.
I shall rather disable that stuff until
somebody finds out what it does. (ECMA-130
talks of Reed-Solomon, GF(2^8), ...
ec.c obviously implements an optimized
algorithm based on function value tables.)
Have a nice day :)