[ On Saturday, November 9, 2002 at 22:23:05 (+0100), Julio Merino wrote: ]
> Subject: pkg/18997: textproc/crimson fails to extract with pax
>
> >How-To-Repeat:
> 	cd /usr/pkgsrc/textproc/crimson
> 	make fetch
> 	( use pax to extract the fetched file ) -> fails

Manually fetching and uncompressing the file and then looking at it
reveals:

$ file *.tar     
crimson-1.3.3-bin.tar: POSIX tar archive

I.e. it was not created by GNU Tar (even though GNU Tar can successfully
read it....)  (file says "GNU Tar archive" for GNU Tar archives :-)

$ pax -vOf *.tar
[[ .... ]]
-rw-r--r--  1 root     wheel       5626 Nov  1  2001 crimson-1.1.3/examples/index.html
-rw-r--r--  1 root     wheel       2070 Nov  1  2001 crimson-1.1.3/README.html
pax: Invalid header, starting valid header search.
-rw-r--r--  1 root     wheel       5626 Nov  1  2001 crimson-1.1.3/examples/index.html
pax: End of archive volume 1 reached
pax: ustar vol 1, 183 files, 3061760 bytes read, 0 bytes written.
$ 
$ gtar -tvf *.tar
[[ .... ]]
-rw-r--r-- 0/0            5626 2001-11-01 01:26 crimson-1.1.3/examples/index.html
-rw-r--r-- 0/0            2070 2001-11-01 01:26 crimson-1.1.3/README.html
$ 

Examination of the tail end of the archive shows that garbage follows
the trailer record on the README.html file, and after that another file
header and following that yet another almost (but not quite) complete
copy of the second last file.

I'm not sure anything needs to be done, or can be safely done, to fix
this (I'm not sure what heuristic GNU Tar could be using -- it should
report an error too!).  The bug is in whatever created this archive.

-- 
								Greg A. Woods

+1 416 218-0098;            <g.a.woods@ieee.org>;           <woods@robohack.ca>
Planix, Inc. <woods@planix.com>; VE3TCP; Secrets of the Weird <woods@weird.com>