Page 6 of 7 FirstFirst ... 4567 LastLast
Results 151 to 180 of 201

Thread: LZA archiver

  1. #151
    Expert
    Matt Mahoney's Avatar
    Join Date
    May 2008
    Location
    Melbourne, Florida, USA
    Posts
    3,255
    Thanks
    306
    Thanked 778 Times in 485 Posts
    Updated 10GB (systems 4 and 7) with lza 0.70b. Still need to do LTCB. http://mattmahoney.net/dc/10gb.html

    Edit: done. Moves up 3 places http://mattmahoney.net/dc/text.html
    Last edited by Matt Mahoney; 20th November 2014 at 23:12.

  2. The Following User Says Thank You to Matt Mahoney For This Useful Post:

    Nania Francesco (20th November 2014)

  3. #152
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    At the moment I'm working on the DLL that will allow anyone to use LZA compression with greater ease and making it independent of the language used for compilation. Of course when I'm finished it will give specifications for use with the prototypes of the commands. I'll release a Windows version of LZA that uses the DLL to compress and manage the LZA archives.

  4. The Following 2 Users Say Thank You to Nania Francesco For This Useful Post:

    Bulat Ziganshin (10th January 2015),load (23rd November 2014)

  5. #153
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    I've been working for more than a month to the windows version of LZA which is really interesting in my opinion. Allows you to manage the archives, read content, info, check the database, decide not to check the archive during extraction and much more. Compressed programs launched do not give any problem being unpacked into a temporary folder. We release an unofficial trial.

    There is much to do and little time. I hope it's interesting!

    Added LZA.x86.exe version! LZAwin x86 version require more time !
    Attached Files Attached Files
    Last edited by Nania Francesco; 15th January 2015 at 09:25.

  6. The Following 3 Users Say Thank You to Nania Francesco For This Useful Post:

    Bulat Ziganshin (10th January 2015),Jarek (10th January 2015),load (11th January 2015)

  7. #154
    Member FatBit's Avatar
    Join Date
    Jan 2012
    Location
    Prague, CZ
    Posts
    189
    Thanks
    0
    Thanked 36 Times in 27 Posts
    Dear Mr. Nania,

    LZAWin.exe => system error 193, text description is not available
    LZA.exe => is not valid win32 application

    All on Win XP SP3 32b Czech edition.

    Best regards, FatBit

  8. #155
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Dear Mr. Nania,

    LZAWin.exe => system error 193, text description is not available
    LZA.exe => is not valid win32 application

    All on Win XP SP3 32b Czech edition.
    Best regards, FatBit

    This test unofficial version is for x64 core's!
    I'm trying to run version x86 or 32b (both DLL and LZAWin)
    Last edited by Nania Francesco; 11th January 2015 at 18:57.

  9. #156
    Expert
    Matt Mahoney's Avatar
    Join Date
    May 2008
    Location
    Melbourne, Florida, USA
    Posts
    3,255
    Thanks
    306
    Thanked 778 Times in 485 Posts
    A little better compression but also a little slower than v0.70b. http://mattmahoney.net/dc/10gb.html

  10. The Following User Says Thank You to Matt Mahoney For This Useful Post:

    Nania Francesco (12th January 2015)

  11. #157
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Thanks Matt for the test! It's a little slower decompression as each file verification (crc32). In the next release I'll put the possibility of excluding the verification as for the DLL.
    The compression is slow because I changed search code match!
    Compiled with gcc 4.9.2 (x64)

  12. #158
    Member
    Join Date
    Nov 2013
    Location
    Kraków, Poland
    Posts
    682
    Thanks
    215
    Thanked 212 Times in 131 Posts
    You can include checksum in ANS to remove the speed penalty: fix the initial state of encoding and if there was an error, you will get a random final value while decoding instead - using 32bit rANS this way should be comparable with crc32.

  13. The Following User Says Thank You to Jarek For This Useful Post:

    Nania Francesco (12th January 2015)

  14. #159
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    This your solution even if referred to a piece of code compressed (averaging over 60-70% the same) might be an interesting test solution.

  15. #160
    Programmer Bulat Ziganshin's Avatar
    Join Date
    Mar 2007
    Location
    Uzbekistan
    Posts
    4,497
    Thanks
    733
    Thanked 660 Times in 354 Posts
    crc32 can be checked with speed of 1 byte/cycle, i.e. 4GB/s per one 4GHz core

    CRC32C has built-in CPU support since 2008 (intel) and 2011 (AMD), processing 4 bytes/cycle in x86 code, and 8 bytes/cycle in x64 code

    in both cases, we should deal with cpu command delays by processing multiple buffers simultaneously and then combining their CRCs using well-known math formulas. look f.e. at google crc library

    it's also possible to use cryptographic hashes that have multi-gigabyte/second speeds

  16. #161
    Member
    Join Date
    Nov 2013
    Location
    Kraków, Poland
    Posts
    682
    Thanks
    215
    Thanked 212 Times in 131 Posts
    If the compression ratio is the only priority, you can store a few symbols in the initial state of ANS.
    But you can also use it as a checksum instead (there are also intermediate options available: e.g. initial_state = L + first_symbol, too large state means error).
    So flipping one bit in the file to decode would result in change of the state value after corresponding step by some small power of 2.
    The question is if you could cancel the modification of state it made by flipping some other bits?
    Even if the modification wouldn't change the symbol produced in this step, the bits read in the next step choose different positions (the modification is shifted to more significant bits) and so cannot cancel the change.
    The probability of accidentally getting the proper checksum with violated stream for 32bit rANS and crc32 seem to be comparable.

  17. #162
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Added LZA.x86.exe version! LZAwin x86 version require more time !

  18. #163
    Member kbollmann's Avatar
    Join Date
    Jan 2015
    Location
    Georgetown
    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts
    for those that used to use LZC10.exe on old 16bit machines and still need to maintain LZC compression and like a command line LZC10 compressor for 32 and 64 bit windows systems we have created LZC10R.exe it also automatically can create a .asm text file that can be used in a number of assemblers with a header that contains compressed and uncompressed length.
     
    [TR]
    [TD="width: 544"]http://store.esellerate.net/s.aspx?s=STR8168998046
    [/TD]
    [/TR]

  19. #164
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Released last version 0.81 64bit version !
    - official windows version (at moment option password is not active and max single file size is limited to 33 GB)
    - new archive (no crc32) with internal error control !
    - new fast compression !

    download from:
    http://heartofcomp.altervista.org/

  20. #165
    Member
    Join Date
    May 2012
    Location
    United States
    Posts
    323
    Thanks
    178
    Thanked 52 Times in 37 Posts
    Tested LZAWin:

    SFC
    Code:
    Default :  (14,248,539 
    Best :     (13,318,642
    Good job! LZAWin needs SFX and it will be a very practical archiver for everyday use!

    Very good Nania!

  21. #166
    Member
    Join Date
    Aug 2014
    Location
    Argentina
    Posts
    468
    Thanks
    203
    Thanked 81 Times in 61 Posts
    Thanks! What about x86? I'm not able to test x64...

  22. #167
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Thanks! What about x86? I'm not able to test x64...
    Added LZAx86.exe Dos version !

    download from:
    http://heartofcomp.altervista.org/


    The problem that I'm not releasing the x 86 version of Windows LZAwin and I use another programming language to create this version and DLL (written in c++ and compiled with gcc) x 86 version often goes in crush!
    Last edited by Nania Francesco; 27th January 2015 at 12:47.

  23. #168
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    I have a question to ask all you experts with regard to the control of error (sorry if the question is made again).

    Using compression narANS I am fortunate to have a signal in case of error. In fact at the end of the decompression value known as rans must always be equal to the same number called narans_interval. However, this involves theoretically and in my opinion that this can be considered only a test of the success of compression but not of equivalence between what you extracted and the original file. This means that I will have to put an additional check crc32 for those who want to be more sure of the integrity to 100%?
    Last edited by Nania Francesco; 27th January 2015 at 13:03.

  24. #169
    Programmer Bulat Ziganshin's Avatar
    Join Date
    Mar 2007
    Location
    Uzbekistan
    Posts
    4,497
    Thanks
    733
    Thanked 660 Times in 354 Posts
    Nania, look at the http://encode.su/threads/2114-Superf...hing-algorithm thread. CRC32 is definitely not the fastest nor the secure algorithm

  25. #170
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    @ Bulat
    but in your opinion so requires an additional control then?

  26. #171
    Programmer Bulat Ziganshin's Avatar
    Join Date
    Mar 2007
    Location
    Uzbekistan
    Posts
    4,497
    Thanks
    733
    Thanked 660 Times in 354 Posts
    i don't understand you. if you ask me about quality of ANS built-in check, i don't know anything about it

  27. #172
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Understood. You which system would you recommend so UMAC?

  28. #173
    Member
    Join Date
    Nov 2013
    Location
    Kraków, Poland
    Posts
    682
    Thanks
    215
    Thanked 212 Times in 131 Posts
    Nania, you will never get 100% integrity guarantee with hash - there is always a tiny probability that you will accidentally get the proper checksum with improper file - about 2^-32 for crc32 (less than 1 per billion).
    In ANS you have the same situation - with some small probability, you will accidentally get the proper state. For 32bit rANS this probability is ~2^-31.95 (a bit worse due to nonuniform probability distribution of states).

  29. #174
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    @Jarek
    Actually it is as you say Jarek. I was thinking that I could use the fixed State rans_interval even to insert a password security but not for encryption.

  30. #175
    Member
    Join Date
    Nov 2013
    Location
    Kraków, Poland
    Posts
    682
    Thanks
    215
    Thanked 212 Times in 131 Posts
    What do you mean by a difference between "password security" and encryption?
    If you want to include encryption, you should rather use tANS not rANS: e.g. https://dl.dropboxusercontent.com/u/...ANSexample.png
    If you want to use obtained checksum for security purposes, finding collisions does not seem a simple task, but I would be careful.
    Generally security applications of ANS require further research.

  31. #176
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    I thought of doing this.
    1) enter a password too long 256 bit (max 32 char)
    2) create a 32-bit Hash of the 256 bit Hash rans_interval referred to in paragraph 1 using Xor (rans_interval compression will not be touched of course) that will be recorded.
    3) On entering the password and decompression with a reverse process if you will get rans_interval you can extract!
    It seems workable.

  32. #177
    Member
    Join Date
    Nov 2013
    Location
    Kraków, Poland
    Posts
    682
    Thanks
    215
    Thanked 212 Times in 131 Posts
    Effectively you get 32bit password for each block this way, which could be easily e.g. brute-forced.
    If you want a decent encryption, the coding step should depend on the key - what can be easily achieved in tANS.
    In rANS you could apply a key-dependent permutation of alphabet, or some small perturbation to the quantizer ... but for better protection you should rather switch to tANS.

  33. #178
    Programmer Bulat Ziganshin's Avatar
    Join Date
    Mar 2007
    Location
    Uzbekistan
    Posts
    4,497
    Thanks
    733
    Thanked 660 Times in 354 Posts
    for an ecryption you may use library such as http://www.gladman.me.uk/cryptograph...y/fileencrypt/ . i have studied rar/winzip/7zip approaches and implemented the same method yourself. i don't think that users will trust any handmade solution (i.e not AES finalist or so)

  34. #179
    Expert
    Matt Mahoney's Avatar
    Join Date
    May 2008
    Location
    Melbourne, Florida, USA
    Posts
    3,255
    Thanks
    306
    Thanked 778 Times in 485 Posts
    I agree. If you want encryption then use AES. If you want to invent your own encryption algorithm, then I suggest you get advice from an expert because amateur attempts are almost always broken. I also suggest you thoroughly document the encryption algorithm and publish source code if you want people to trust it.

  35. #180
    Tester
    Nania Francesco's Avatar
    Join Date
    May 2008
    Location
    Italy
    Posts
    1,565
    Thanks
    220
    Thanked 146 Times in 83 Posts
    Thanks Matt I guess I'll use AES, actually I do not want to engage in something that I don't know well.

Page 6 of 7 FirstFirst ... 4567 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •