Re: tar -cvfX save.tar foo ./dirtosave/..

From: Joerg Schilling (js_at_cs.tu-berlin.de)
Date: 08/28/03 

Date: 28 Aug 2003 21:25:24 GMT

In article <pan.2003.08.28.16.01.00.64876@socal.rr.com>,
Ed Murphy <emurphy42@socal.rr.com> wrote:
>On Thu, 28 Aug 2003 13:09:29 +0200, Peter T. Breuer wrote:
>
>> Uh - I've used gnu tar on msdos without problems! And on solaris! And
>> on irix. And on netbsd. And .... oh, forget it. I don't feel trapped.
>
>Well, sure: GNU tar on Solaris ought to understand the GNU-and-not-POSIX
>extensions produced by GNU tar on Linux.

Well, you can ride a bike with square wheels if you manage to built the right
roads before. But why do this when the world did agree on circkled wheels?
All overall there are planar roads...

GNU tar is like a bike with square wheels in a world of planar roads....

>Since Joerg is (again) too apoplectic to cut to the heart of the matter,
>I suppose I'll give it another shot:

Do you have a way to deal with a troll like Mr. Breuer who says:

        convince me but I will neither believe your writings nor
        try to get independant information that could prove that I am wrong?

He refuses to read standards that prove me, he refuses to do simple tests
that prove me, ....

He constantly repeats his junk and asks: tell me why this is that way!
But he ignores that I did answer his questions several times before.

He is just an evil-minded troll.

>1) If you use GNU tar on Linux to tar a file whose full-path is >100
> characters, then use tar (not GNU tar, and not star -Hgnutar) on
> some other OS to untar it, then is the file untarred with its
> correct >100-character full-path? (Joerg's evidence strongly
> implies that the answer is "no".)

Correct, this may be easily proved by anybody. But Mr. Breuer refuses
to do it although he stated that he has access to more than Linux ony.

>2) Is the file untarred with the correct contents, but a truncated
> path, and accompanied by a "@LongLink" file that states what the
> full-path is supposed to be? (Joerg's evidence seems to imply
> that the answer is "yes".)

If there is only one such file in the archive you are right.
Otherwise the file ././@LongLink gets overwritten several times (if you
are root) or you get a permission denied message (if the tar program
writes error messages).

>3) Joerg seems to believe that, instead of the behavior noted in 2),
> that GNU tar with --posix should either use POSIX extensions
> (which, mind you, sufficiently old tar programs won't understand
> either) or simply refuse to tar the file.

You need to use a really old tar (made before 1993) if you like to get
into problems with missing POSIX compliance.

> a) If the POSIX extensions are used, then what happens if you
> try to untar that with an old tar program that doesn't
> understand the POSIX extensions?

file types like named pipes and bdev/cdev are extracted as plain file
of size 0.

Files with long pathnames will without a name dependant path prefix.

> b) Is the behavior in a) (whatever it is) superior to the behavior
> in 2)?

yes because it is not worse and there are much more tar implementations that do
support POSIX.

> c) Is refusal to tar the file superior to the behavior in 2)?

Yes, because this warns you about incompatibilities before they may occur.

>4) Can someone track down the history of the POSIX specs that Joerg
> holds up as examples of GNU's failure? Specifically, can someone
> identify when they exited the draft phase and became official? (I
> am not too worried about this point, though, as GNU tar itself
> gives credence to certain specs by saying "we're not compatible
> with them yet".)

I did write it many times and it is easy to prove: the oldest final
version is from 1988. So this is before GNU started to make GNU tar
from PD tar and before GNU tar introduced non POSIX things.

        PD tar did not contradict POSIX.

>5) Can someone track down the history of GNU tar? Specifically, can
> someone identify any past dates on which the current claims ("we're
> incompatible with POSIX such-and-such but we're working on it")
> already existed, with explicit evidence? Basically, let's find out
> how old those claims really are.

The big problem is that the false claims in the tar.info file that Mr.
Breuer constantly quotes have been written in 1994 by people wo most likely
nerver had any relationship to the people who did introduce the non-compliances
in 1989.

Many of my claims may be easily proved if you just get the PD tar (SUG-tar)
source I did keep since december 1987.

see: ftp://ftp.berlios.de/pub/star

The FSF people who did write the false claims may not even have written the lies
by intent. It may be that they just did a lousy research job before tey did
start writing.

The real bad thing is that they did refuse to correct the text when I told them
that there is something wrong in 1994. I also did tell them that they may get
all may files that allwow to find out the truth, but they have not been
interested to correct their wrong claims :-( 

-- 
EMail:joerg@schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
      js@cs.tu-berlin.de		(uni)  If you don't have iso-8859-1
      schilling@fokus.fraunhofer.de	(work) chars I am J"org Schilling
URL:  http://www.fokus.fraunhofer.de/usr/schilling ftp://ftp.berlios.de/pub/schily

Relevant Pages