[r-t] archives & zipping

[Ian McCulloch]

On Tue, 20 May 2014, Robin Woolley wrote:

[...]

> I guess whoever gzips the file does too much directory recursion. Useful for 
> backing up, etc, lots of files but a little inappropriate for one file! 
> However, this does not explain the original change.

I don't know what has changed, but it isn't really a problem with the 
archive.  The gzip file format, described at 
http://www.gzip.org/zlib/rfc-gzip.html contains an optional header for the 
original filename.  For the bellringers.net archives, this is of the form, 
eg,

/usr/local/cpanel/3rdparty/mailman/archives/private/ringing-theory_bellringers.net/2014-May.txt

I don't know which version of gzip uses the full path, the .gz files I 
create with GNU gzip only contain the filename itself without the 
directories, and I can't see how to change that.  But in any event, the 
filename format in the archives has been in the same full path format 
since at least 2004.

But it is rather crazy behaviour of your gzip utility to require you to 
navigate through the complete path to the file.  The gzip format isn't an 
'achive', it only contains a single file.  None of the graphical or 
command-line tools I have even give a hint that there is a full path 
embedded in the optional header - I used a binary viewer to tell that it 
existed.

Cheers,
Ian