jw, Mon 2. Dez 00:24:27 CET 2013

Implement least surprise:
 Option -1 should be default. 
 A new option -r should make it recursive.

				jw, Tue Jun 18 15:09:19 CEST 2013

Feature: support *.phar files. Example: install-pear-nozlib.phar from package php5.
This example wants write access to /usr/share/php5/PEAR/*
No idea... (Forget fakeroot. It fakes a root user, not a root filesystem)

				jw, Thu Mar 14 19:51:33 CET 2013

file_unpack -L logfile.log .
would list the logfile within itself. Avoid this!

				jw, Thu Jan 24 11:39:37 CET 2013

According to: https://rt.cpan.org/Ticket/Display.html?id=73670#txn-1039064
With multipart rar archives we extract the same thing multiple times.
A test archive can be created with: rar a -v10M test.rar *.webm
  - post process the result tree, and see if there are multiple indentical
    (colliding!) subtrees. That would be quite generic, but not so efficient.
  - preprocess the input files, so that an ignore list prevents us from
    touching the other parts again. This would need rar specific code, which I
    am not so happy about. Maybe call rar in a tv mode first, and learn from there, 
    all the input filenames used?
  - run /usr/bin/unrar x -o- -p- -iul -kb -y test.part03.rar
    *without -n* so that it prints messages while working. The messages look like this:
       Extracting from test.part04.rar
       ...         design-patterns-in-ruby-20120426.webm                     34%
       Extracting from test.part05.rar
       ...         design-patterns-in-ruby-20120426.webm                     43%
       Extracting from test.part06.rar
       All OK

    Parse this for qr{^Extracting from (.*)$} and blacklist all filenames found with $1
    The protocol for mimetype-handlers needs to be extended so that we can safely return 
    the list of unpacked files.

				jw, Wed May  4 11:58:27 CEST 2011

file_unpack /space/legaldb_nfs/auto-co/api.opensuse.org/mozilla\:Factory/js%1.8.5%r3/
this pulls 1.7gb incore. why?

				jw, Mon Mar 28 21:01:45 CEST 2011

lib/ldig/fs.pm had this on its sub get_file_mimetype(): "file -i -b %s"
# file returns things like:
# 'text/plain; charset=us-ascii'
# 'application/x-tar, POSIX'
# 'text/x-c'
# we truncate at the , or ;
# If $fname ends in '|' we assume it is a command like 'gzip -dc file.gz|'
# vim-7.0-extra.tar.bz2/vim70/csdpmi4b.zip~ldig/BIN/CWSDPMI.DOC
# is a text file, but file says: application/x-shockwave-flash
# vim70/runtime/macros/maze/poster: message/news\0118bit
# FIXME: DELTA rpms have DLT3 magic after unrpm. How do we handle this?

				jw, Wed Dec 15 20:31:41 CET 2010

Browse through this list of formats
and see what we can do.

				jw, Wed Nov 24 18:15:15 CET 2010

 try unshield for .cab files that do not extract with cabextract.
 Those would be Installshield CAB files rather than Microsoft CAB files.

				jw, 2010-09-26 20:21

How to decode Metamail contents?
 The firt part is usually html, and thus file -i says text/html
 ignoring the mime headers.
-> file_unpack -v -m 
  shows the multipart; header in the third filed.

 Content-Type: multipart/alternative;
 Content-Type: multipart/
 Content-Disposition: attachment; filename=Unknown.jpeg;

metamail -w -x < saved-mail-body-from-mutt
## often says says
# metamail: Incomplete multipart message -- unexpected EOF
# Cannot handle any part of multipart/alternative message

-> use perl-MIME-tools, this is more flexible.

# when I extract the parts between the boundary string and feed this
# to metamail one by one, it works well.
env METAMAIL_TMPDIR=. metamail -d -w jpeg.mm
 This message contains 'image/jpeg`-format data.

 which is being decoded and written to the file named "./Unknown.jpeg".
 If you do not want this data, you probably should delete that file.
 Wrote file ./Unknown.jpeg