Darren New wrote:
> Warp wrote:
>>>> Actually, it sucks pretty badly for manipulating actual tape archives,
>>>> when you get right down to it.
>>
>>   The tar *program* maybe, but manipulating the tar *format* is just a
>> question of selecting the program you like. 
> 
> I was speaking of actual tape archives on actual tapes.

Actually, the format is pretty sucky for tape archives too.

There's no central directory of files in the archive, so you have to 
read the entire tape to determine what's on it.

There's no archive-level metadata, so you can't tell when the tape was 
created, or why, or how many tapes are in the set, or what parameters 
might be needed to extract it (see the "B" option, for example).

There metadata is fixed-format, which means it's incompatible with newer 
versions of the format, because there isn't even a place to record what 
version of the metadata is there.

The metadata is very single-machine specific, so (for example) you can't 
restore the tape to a different machine with a different username->uid 
mapping and expect it to work. Indeed, you can't even tell what the 
uid's represent by inspection so you can't even write a program to 
demangle it when it's done.

The data is all stored in block sizes, wasting a lot of space in spite 
of the fact that tar cannot update the middle of an archive nor does the 
tar program seek the file even if it's on disk. This also causes one to 
have to read and write in block sizes, even tho the file is "supposed" 
to be an "array of bytes" kind of thing.

Just as a few nits... :-)

-- 
Darren New / San Diego, CA, USA (PST)
  Ever notice how people in a zombie movie never already know how to
  kill zombies? Ask 100 random people in America how to kill someone
  who has reanimated from the dead in a secret viral weapons lab,
  and how many do you think already know you need a head-shot?

Post a reply to this message