Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.compression > #1957 > unrolled thread

Compression Update: Preliminary Good News

Back to article view | Back to comp.compression

Contents

  Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-05-27 06:32 -0700
    Re: Compression Update: Preliminary Good News Ernst <Ernst_Berg@sbcglobal.net> - 2013-05-29 16:40 -0700
      Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-09 15:10 -0700
        Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-10 08:22 -0700
    Re: Compression Update: Preliminary Good News Ernst <Ernst_Berg@sbcglobal.net> - 2013-06-17 11:23 -0700
      Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-18 07:15 -0700
        Re: Compression Update: Preliminary Good News "George Johnson" <matrix29@charter.net> - 2013-06-19 00:23 -0400
          Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-19 07:08 -0700
          Re: Compression Update: Preliminary Good News Phil Carmody <thefatphil_demunged@yahoo.co.uk> - 2013-06-27 10:01 +0300
            Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-27 06:05 -0700
              Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-07-07 10:32 -0700
                Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-07-19 11:04 -0700
    Re: Compression Update: Preliminary Good News Harry Potter <rose.joseph12@yahoo.com> - 2013-06-26 06:14 -0700

#1957 — Compression Update: Preliminary Good News

I'm still working on file compression, and right now, the numbers are good--for text files.  They are usually better than by the deflate technique--even if by 1-2%.  However, binary files are poorly compressible, and sometimes, the coder locks up in the middle of the coder process.  The last happened to me before, and it was due to improper handling of unsigned chars.  Now, I have to optimize the coder and implement the decoder and see if it stands.

[toc] | [next] | [standalone]

#1961

On Monday, May 27, 2013 6:32:02 AM UTC-7, Harry Potter wrote:
> I'm still working on file compression, and right now, the numbers are good--for text files.  They are usually better than by the deflate technique--even if by 1-2%.  However, binary files are poorly compressible, and sometimes, the coder locks up in the middle of the coder process.  The last happened to me before, and it was due to improper handling of unsigned chars.  Now, I have to optimize the coder and implement the decoder and see if it stands.

Good Luck Man.

[toc] | [prev] | [next] | [standalone]

#1971

On Wednesday, May 29, 2013 7:40:41 PM UTC-4, Ernst wrote:
> Good Luck Man.

Thank you.  I achieved the numbers mentioned after implementing Adaptive Huffman codes on an 8-bit codec--but then I broke it and lost the working properly.  I've never been able to fix it.  :(  However, I believe I'm very close to Win32 zipped folders' compression ratios.  The 16-bit version is off the compression ratio by .8% on the test file, and the 8-bit, by a few bytes on its test file.  Both files are highly compressible.  If I apply Adaptive Huffman codes, I should do well.  Thank you for listening.

[toc] | [prev] | [next] | [standalone]

#1972

Good news!  I found a working copy and tweaked it.  It works now.  :)  It seems to do better than Win32 compressed folders on text files.  Unfortunately, it locks up on binary files.  :(

[toc] | [prev] | [next] | [standalone]

#1975

I know the routine of coding and proving. It's a skill we learn over time.

[toc] | [prev] | [next] | [standalone]

#1976

On Monday, June 17, 2013 2:23:28 PM UTC-4, Ernst wrote:
> I know the routine of coding and proving. It's a skill we learn over time.

Bad news!  :(  It didn't work on binary files--just text files.  I'm working on another copy now, and it's *not* showing promise.  The numbers are *way* off.  :(

[toc] | [prev] | [next] | [standalone]

#1977

"Harry Potter" <rose.joseph12@yahoo.com> wrote in message 
news:f94d1526-af5c-4ad4-917a-adc1c53adc7e@googlegroups.com...
> On Monday, June 17, 2013 2:23:28 PM UTC-4, Ernst wrote:
>> I know the routine of coding and proving. It's a skill we learn over 
>> time.
>
> Bad news!  :(  It didn't work on binary files--just text files.  I'm 
> working on another copy now, and it's *not* showing promise.  The numbers 
> are *way* off.  :(

    No biggie, there is still plenty of demand for text-centric data 
compression.
    It's like an inventor saying that they've got a new data compression 
routine that works great on 3D game objects and then being depressed that 
the new compression routine fails at compressing bitmap images or video or 
HTML files.

    Just because it isn't useful to the wider audience of general consumers 
or corporate interest, does not mean by itself, that it is useless.
    Sometimes it just takes a bit of imaginative Lateral Thinking to best 
utilize a new invention that nobody else appreciates.
    Otherwise the classic businessman's trick is to take a somewhat useless 
product and then aggressive create a market for by initiating market demand 
triggers. 

[toc] | [prev] | [next] | [standalone]

#1978

George Johnson: I am glad for your response, but I will stubbornly stick to a general-prupose technique, as the ideology behind my ideas should work on almost *any* file.

[toc] | [prev] | [next] | [standalone]

#1991

"George Johnson" <matrix29@charter.net> writes:
> "Harry Potter" <rose.joseph12@yahoo.com> wrote in message 
> news:f94d1526-af5c-4ad4-917a-adc1c53adc7e@googlegroups.com...
> > On Monday, June 17, 2013 2:23:28 PM UTC-4, Ernst wrote:
> >> I know the routine of coding and proving. It's a skill we learn over 
> >> time.
> >
> > Bad news!  :(  It didn't work on binary files--just text files.  I'm 
> > working on another copy now, and it's *not* showing promise.  The numbers 
> > are *way* off.  :(
> 
>     No biggie, there is still plenty of demand for text-centric data 
> compression.
>     It's like an inventor saying that they've got a new data compression 
> routine that works great on 3D game objects and then being depressed that 
> the new compression routine fails at compressing bitmap images or video or 
> HTML files.
>     Just because it isn't useful to the wider audience of general consumers 
> or corporate interest, does not mean by itself, that it is useless.

Such domain-specific algorithms are if anything more interesting than
general purpose ones, as in general, you know what domain your data
stream belongs too. Not utilising that additional knowledge is waste.

Phil
-- 
If "law-abiding citizens have nothing to fear" from privacy-invading 
technologies and policies, then law-abiding governments should have
nothing to fear from whistleblowers.

[toc] | [prev] | [next] | [standalone]

#1993

On Thursday, June 27, 2013 3:01:39 AM UTC-4, Phil Carmody wrote:
> Such domain-specific algorithms are if anything more interesting than general purpose ones, as in general, you know what domain your data stream belongs too. Not utilising that additional knowledge is waste.

You're right.  I also have ideas to better text files and files already compressed using the Deflate technique.  But for now, I want to create something general.  :)

[toc] | [prev] | [next] | [standalone]

#1995

New update: The problem is that I forgot to swap the part of each node in Adaptive Huffman codes containing a node's symbol.  Now it works on both text and binary files, and it does *alot* better than Deflate.  :)  The code has a custom-compilation feature to give the compression ratio, but when I enable this, the compressed size of the file jumps considerably.  It shouldn't do this as only the main module, which doesn't really handle the compression--it just reads the file into memory.  :(  I might reveal a piece of this code, but, as it's written for an 8-bit computer, it might not help here.

BTW, I plan to eventually attack more modern platforms later.  ;)

[toc] | [prev] | [next] | [standalone]

#2006

I fixed the bug and made other bug fixes and other edits, and now the ratio seems almost too good to be true!  :)  Moreover, the result is consistent: if done properly, the results are consistently and considerally better than with the deflate technique.  Now to optimize and debug....

[toc] | [prev] | [next] | [standalone]

#1990

I seem to have finally gotten the codec--the coder, anyway--to work.  The numbers are good, but before I can be sure, I have to implement the decoder and debug both.  First, I'm gonna optimise the coder so that it'll both fit on an 8k ROM image for the Commodore 64 and be fast enough for any 8-bit computer.  Thank you for listening.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.compression

csiph-web