Re: 7-Zip / Bzip2

2008-05-06 Thread Samuel J. Greear


Samuel J. Greear [EMAIL PROTECTED] wrote in message 
news:[EMAIL PROTECTED]


Matthew Dillon [EMAIL PROTECTED] wrote in message 
news:[EMAIL PROTECTED]


:Hi,
:
:Posted this to kernel@ by accident, please reply here instead :)
:
:I just wanted to know if there's any interest for the devs to add
:something like p7zip to the base install; even if it's a simple fork
:that only supports 7z.  While 7zip is about as obnoxiously slow as
:bzip2, it usually gets much better compression.
:
:That's not why I'm suggesting it though - what really gets me is that
:bzip2 has no list option.  Does that 10 gb bzip2 backup archive
:contain 100gb of data, or 200gb?  Other than dumping the entire
:archive to /dev/null through wc, there's really no way to do it.  Gzip
:will list files, but its compression ratio is awful.
:
:I imagine that other OSes are going to be watching Dragonfly very
:carefully in the next while as new the features (especially HAMMER)
:mature.  Maybe adding 7z will get yet another bandwagon going and
:there will be support across the board :)
:
:Best Regards,
:Ben Cadieux

   Well, I think not in base, at least not unless a lot of people
   are using it.  p7zip is readily available via the pkgsrc tree
   and that's the most reasonable method of accessibility for
   now.

-Matt
Matthew Dillon
[EMAIL PROTECTED]


I do not know if DragonFly is actively tracking libarchive, but it seems
to be the ticket for implementing new archiving/compression methods
through a common mechanism. If one wanted to see 7z functionality
in base implementing a libarchive provider is probably the way to go.

I did a quick test at one point on the dfly distribution ISO and as I
recall 7z was ~60% the size of bzip2 using standard settings.
food+thought, etc.

Sam


I just noticed I still have these sitting here so I figured I would post
sizes. 1.12.1 ISO 294M, bz2 108M, 7z 74M, zip 118M, gz 120M.

Sam 



Re: 7-Zip / Bzip2

2008-05-06 Thread Samuel J. Greear


Samuel J. Greear [EMAIL PROTECTED] wrote in message 
news:[EMAIL PROTECTED]


Matthew Dillon [EMAIL PROTECTED] wrote in message 
news:[EMAIL PROTECTED]


:Hi,
:
:Posted this to kernel@ by accident, please reply here instead :)
:
:I just wanted to know if there's any interest for the devs to add
:something like p7zip to the base install; even if it's a simple fork
:that only supports 7z.  While 7zip is about as obnoxiously slow as
:bzip2, it usually gets much better compression.
:
:That's not why I'm suggesting it though - what really gets me is that
:bzip2 has no list option.  Does that 10 gb bzip2 backup archive
:contain 100gb of data, or 200gb?  Other than dumping the entire
:archive to /dev/null through wc, there's really no way to do it.  Gzip
:will list files, but its compression ratio is awful.
:
:I imagine that other OSes are going to be watching Dragonfly very
:carefully in the next while as new the features (especially HAMMER)
:mature.  Maybe adding 7z will get yet another bandwagon going and
:there will be support across the board :)
:
:Best Regards,
:Ben Cadieux

   Well, I think not in base, at least not unless a lot of people
   are using it.  p7zip is readily available via the pkgsrc tree
   and that's the most reasonable method of accessibility for
   now.

-Matt
Matthew Dillon
[EMAIL PROTECTED]


I do not know if DragonFly is actively tracking libarchive, but it seems
to be the ticket for implementing new archiving/compression methods
through a common mechanism. If one wanted to see 7z functionality
in base implementing a libarchive provider is probably the way to go.

I did a quick test at one point on the dfly distribution ISO and as I
recall 7z was ~60% the size of bzip2 using standard settings.
food+thought, etc.

Sam


I got yelled at on IRC for not providing times, so here:

Key:
   Selected Archiver - Compression Time - Threads - FileSize

Source file:
   dfly-1.12.2_REL.iso - 295,512KB

Bzip2 - 0:27 - 4 - 108,742KB
Bzip2 - 1:24 - 1 - 108,742KB
Decompress: 0:10

GZip - 0:55 - 1 - 118,024KB
Decompress: 0:10

Zip - 0:55 - 4 - 118,024KB
Zip - 0:55 - 1 - 118,024KB
Decompress: 0:11

7Zip - 1:50 - 2 - 73,952KB
7Zip - 3:09 - 1 - 73,952KB
Decompress: 0:12


All done via the most unscientific methods available (reporting
user time as displayed by the 7Zip user interface), so treat it
as such. Tested on an Intel Q6600 (4x2.4GHz) w/ single SATA
disk.

Default compression settings were used across the board.
Everything was tested using the windows 7-Zip program, which
integrates all of these algorithms.

Sam 



Re: 7-Zip / Bzip2

2008-05-06 Thread Cristi Magherusan
Hello,

On Tue, 2008-05-06 at 09:51 -0700, Samuel J. Greear wrote:
 Samuel J. Greear [EMAIL PROTECTED] wrote in message 
 news:[EMAIL PROTECTED]
 
  Matthew Dillon [EMAIL PROTECTED] wrote in message 
  news:[EMAIL PROTECTED]
 
  :Hi,
  :
  :Posted this to kernel@ by accident, please reply here instead :)
  :
  :I just wanted to know if there's any interest for the devs to add
  :something like p7zip to the base install; even if it's a simple fork
  :that only supports 7z.  While 7zip is about as obnoxiously slow as
  :bzip2, it usually gets much better compression.
  :
  :That's not why I'm suggesting it though - what really gets me is that
  :bzip2 has no list option.  Does that 10 gb bzip2 backup archive
  :contain 100gb of data, or 200gb?  Other than dumping the entire
  :archive to /dev/null through wc, there's really no way to do it.  Gzip
  :will list files, but its compression ratio is awful.
  :
  :I imagine that other OSes are going to be watching Dragonfly very
  :carefully in the next while as new the features (especially HAMMER)
  :mature.  Maybe adding 7z will get yet another bandwagon going and
  :there will be support across the board :)
  :
  :Best Regards,
  :Ben Cadieux
 
 Well, I think not in base, at least not unless a lot of people
 are using it.  p7zip is readily available via the pkgsrc tree
 and that's the most reasonable method of accessibility for
 now.
 
  -Matt
  Matthew Dillon
  [EMAIL PROTECTED]
 
  I do not know if DragonFly is actively tracking libarchive, but it seems
  to be the ticket for implementing new archiving/compression methods
  through a common mechanism. If one wanted to see 7z functionality
  in base implementing a libarchive provider is probably the way to go.
 
  I did a quick test at one point on the dfly distribution ISO and as I
  recall 7z was ~60% the size of bzip2 using standard settings.
  food+thought, etc.
 
  Sam
 
 I got yelled at on IRC for not providing times, so here:
 
 Key:
 Selected Archiver - Compression Time - Threads - FileSize
 
 Source file:
 dfly-1.12.2_REL.iso - 295,512KB
 
 Bzip2 - 0:27 - 4 - 108,742KB
 Bzip2 - 1:24 - 1 - 108,742KB
 Decompress: 0:10
 
 GZip - 0:55 - 1 - 118,024KB
 Decompress: 0:10
 
 Zip - 0:55 - 4 - 118,024KB
 Zip - 0:55 - 1 - 118,024KB
 Decompress: 0:11
 
 7Zip - 1:50 - 2 - 73,952KB
 7Zip - 3:09 - 1 - 73,952KB
 Decompress: 0:12
 
 
 All done via the most unscientific methods available (reporting
 user time as displayed by the 7Zip user interface), so treat it
 as such. Tested on an Intel Q6600 (4x2.4GHz) w/ single SATA
 disk.
 
 Default compression settings were used across the board.
 Everything was tested using the windows 7-Zip program, which
 integrates all of these algorithms.
 
 Sam 
 

The compression time is not an issue for install disks, and
decompression extra time of 7z is insignificant, but the best usage for
this would be when packaging sources, not the installer CD, whose size
will grow slower than the size of CD/DVD/BlueRay... Anyway our CD image
could easily fit a mini-CD, so we don't need this, for now.

Regards, Cristi.

-- 
Cristi Magherusan,
Universitatea Tehnica din Cluj - Napoca
Centrul de Comunicatii Pusztai Kalman
Tel. 0264/401247  http://cc.utcluj.ro


signature.asc
Description: This is a digitally signed message part


Re: 7-Zip / Bzip2

2008-05-06 Thread Matthew Dillon

:I just noticed I still have these sitting here so I figured I would post
:sizes. 1.12.1 ISO 294M, bz2 108M, 7z 74M, zip 118M, gz 120M.
:
:Sam 

Those are impressive numbers.  I will keep an eye on it.  If 7z is
consistently better then bz2 I expect it will become widely adopted.

I just think it is a bit early to put it into base when it is so
readily accessible via pkgsrc.

-Matt
Matthew Dillon 
[EMAIL PROTECTED]


Re: 7-Zip / Bzip2

2008-05-06 Thread Steve O'Hara-Smith
On Tue, 06 May 2008 20:29:53 +0300
Cristi Magherusan [EMAIL PROTECTED] wrote:

 The compression time is not an issue for install disks, and
 decompression extra time of 7z is insignificant, but the best usage for
 this would be when packaging sources, not the installer CD, whose size
 will grow slower than the size of CD/DVD/BlueRay... Anyway our CD image
 could easily fit a mini-CD, so we don't need this, for now.

There is still the little matter of download time and bandwidth
utilisation. Certainly if I saw a 73MB file and a 108MB file with the same
contents I know which one I would choose to download (once I'd assured
myself of being able to unpack it).

-- 
C:WIN  |   Directable Mirror Arrays
The computer obeys and wins.| A better way to focus the sun
You lose and Bill collects. |licences available see
|http://www.sohara.org/


Re: 7-Zip / Bzip2

2008-05-06 Thread Matthew Dillon
Well, the issue is not so much one of download bandwidth as it is one
of accessibility.  Everyone has gzip, not everyone has 7z (yet). 

I would rather not have multiple compression formats on the download
site, nor require that people have 7z in order to be able to unpack
our release CDs!

-Matt
Matthew Dillon 
[EMAIL PROTECTED]


Re: 7-Zip / Bzip2

2008-05-06 Thread Ben Cadieux
 Well, the issue is not so much one of download bandwidth as it is one
 of accessibility.  Everyone has gzip, not everyone has 7z (yet).

 I would rather not have multiple compression formats on the download
 site, nor require that people have 7z in order to be able to unpack
 our release CDs!

This is the 'bandwagon' I was referring to --- it's a chicken and egg
problem.  I can safely say I would rarely be using gzip or bzip2 if 7z
were in the base installs of all Linux/BSD releases.  No one uses
compress any more because gzip/bzip2 are in the base.  If they
weren't, you'd still see .Z files everywhere.

Best Regards,
Ben Cadieux


Re: 7-Zip / Bzip2

2008-05-06 Thread Ben Cadieux
For everyone concerned with timing, I tested compressing a 4gb image
of a quad boot USB key with a variety of junk on it at compression
level 4.

It still beats bzip2 by 90 mb and was almost 3 minutes quicker:

# time -h bzip2 -k key.img
11m59.71s real  10m39.48s user  5.03s sys
# time -h 7z a -mx=4 key.img.7z key.img
8m39.02s real   7m38.57s user   4.95s sys

# ls -la | grep key
-rw-r-   1 sol   wheel  4102029312 Mar  8 11:27 key.img
-rw---   1 root  wheel   904489376 May  6 11:38 key.img.7z
-rw-r-   1 sol   wheel   997632056 Mar  8 11:27 key.img.bz2

Best Regards,
Ben Cadieux


Re: 7-Zip / Bzip2

2008-05-06 Thread Hasso Tepper
Samuel J. Greear wrote:
 I just noticed I still have these sitting here so I figured I would
 post sizes. 1.12.1 ISO 294M, bz2 108M, 7z 74M, zip 118M, gz 120M.

lrz 69M

It doesn't mean that I care though. In fact, I use gzip most of time - 
bandwith is quite cheap nowadays.


-- 
Hasso Tepper