Re: Towards multi-arch: "Multi-Arch: same" file conflicts

["John D. Hendrickson and Sara Darnell" <johnandsara2@cox.net>]

hi.  i'm not an "admin" but.  are you sure compression routines guarantee the same compression for 
the same file?  and when given differing directory names?

I have a hard time wrapping my head around your statement: gzip is "erroneous".

Have fun,  John Hendrickson

look at my quick test ...

# cd /tmp/tmp
# srdiff -v x
STAT:  *li *sp *so =co

receiving incremental file list
x

##-------------------------------------
## HOST: li
## FILE: /tmp/tmp/x
##-------------------------------------
receiving file list ... done
x

sent 42 bytes  received 31838 bytes  63760.00 bytes/sec
total size is 31744  speedup is 1.00
##-------------------------------------
## HOST: sp
## FILE: /tmp/tmp/x
##-------------------------------------
receiving file list ... done
x

sent 42 bytes  received 31838 bytes  63760.00 bytes/sec
total size is 31744  speedup is 1.00
##-------------------------------------
## HOST: so
## FILE: /tmp/tmp/x
##-------------------------------------

# cd ..

# rj gzip -9 tmp/x
STAT:  *li *sp *so =co

## HOST: li
## HOST: sp
## HOST: so
## HOST: co

# rj md5sum tmp/x.gz
STAT:  *li *sp *so =co

## HOST: li
d2b182c7bbc2ce01b1f83d4a2f26b742  tmp/x.gz
## HOST: sp
d2b182c7bbc2ce01b1f83d4a2f26b742  tmp/x.gz
## HOST: so
1f7725bbbd392b6e6276eafd26074965  tmp/x.gz
## HOST: co
7904a5643cb7c318a14497dadc0e2f70  tmp/x.gz

# rj gzip -l tmp/x.gz
STAT:  *link *sparky *sol8 =coco

## HOST: li
         compressed        uncompressed  ratio uncompressed_name
              23955               31744  24.6% tmp/x
## HOST: sp
         compressed        uncompressed  ratio uncompressed_name
              23955               31744  24.6% tmp/x
## HOST: so
         compressed        uncompressed  ratio uncompressed_name
              23955               31744  24.6% tmp/x
## HOST: co
         compressed        uncompressed  ratio uncompressed_name
              23955               31744  24.6% tmp/x


Josselin Mouette wrote:
> Hi,
>
> Le jeudi 17 novembre 2011 à 20:03 +0100, Jakub Wilk a écrit : 
> > http://people.debian.org/~jwilk/multi-arch/same-md5sums.txt
> > http://people.debian.org/~jwilk/multi-arch/same-md5sums.ddlist
> >
> > (If there is no MD5 sum information next to a file name, it means that 
> > the sum for each architecture was different.)
>
> Thanks a lot for this report.
>
> libgnome-keyring and librsvg are fixed in unstable.
> atk1.0, clutter-1.0, libcroco and pango1.0 should all be fixed in svn.
> The error for glib2.0 is looks like a bug in gzip, only two
> architectures are different from the others.
>
> Cheers,