'cpio' utility neede

[Mats Wichmann]

>> 
>> 	If there is anything wrong on the input, you get a
>> 	"phase error" and the thing dies.  If, for any 
>> 	reason, you have to skip one of a multi-media set,
>> 	you get a "phase error" and the thing dies.
> 
> What we need is a utility (or a cpio option) that will continue
> after a phase error, and will attempt to identify the next good
> file.  The bad file could be saved as a /tmp file to allow
> human determination as to its value.  Most of the time, a partial
> recover (noted as such) is preferred to NO recovery.
> 

Yes, CPIO sucketh on all the counts charged. It seems to work
better, on the whole, that most of the other backup stuff
(how many time have you fought errors in dump???), but it suffers
from all of the previously mentioned problems, plus another one:
it cant' tell too much about links, so it archives one copy of
the file for each link when it comes up...(for something like
vi, which is liked to some subset of view, vedit, e, ex, edit;
this is a real serious problem - not everyone has a nine-track tape that
will hold a ton of data). The fixes to CPIO to get around
`out of pahse' errors are quite simple, but not that many people
have source any more (would vendors please provide atg least
this option, if nothing else?????). The business about links
is basically impossible, due to the way CPIO is set up....


The IEEE P1003 (UNIX-like OS's) committee is trying to deal
with this by specifying a more complete but backwards compatible
version of TAR, but besides my not being to pleased with the
specs, AT&T has burned a  lot of bridges in this regard by
supporting CPIO in releases since Sys III came out....

    Mats Wichmann
    Dual Systems
    ...{ucbvax,ihnp4,cbosgd,decwrl,fortune}!dual!mats