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

Groups > comp.compression > #332 > unrolled thread

I want to replace the logic

Back to article view | Back to comp.compression

Contents

  I want to replace the logic jules Gilbert <jules.stocks@gmail.com> - 2011-06-23 16:13 -0700
    Re: I want to replace the logic Noob <root@127.0.0.1> - 2011-06-24 10:40 +0200
      Re: I want to replace the logic jules Gilbert <jules.stocks@gmail.com> - 2011-06-24 06:04 -0700
        Re: I want to replace the logic Earl_Colby_Pottinger <earlcolby.pottinger@sympatico.ca> - 2011-06-24 21:32 -0700
        Re: I want to replace the logic Mark Nelson <snorkelman@gmail.com> - 2011-06-25 11:44 -0700
          Re: I want to replace the logic Captain Obvious <alex.mizrahi@gmail.com> - 2011-06-26 13:39 +0300
            Choosing between GPL and LGPL Noob <root@127.0.0.1> - 2011-06-27 11:31 +0200
            Re: I want to replace the logic BGB <cr88192@hotmail.com> - 2011-06-27 15:11 -0700
              Re: I want to replace the logic jules Gilbert <jules.stocks@gmail.com> - 2011-07-10 14:22 -0700
                Re: I want to replace the logic Sebastian <s.gesemann@gmail.com> - 2011-07-10 23:39 -0700
                Re: I want to replace the logic stan <smoore@exis.net> - 2011-07-11 16:21 -0400
                Re: I want to replace the logic Jim Leonard <mobygamer@gmail.com> - 2011-07-12 08:22 -0700
          Re: I want to replace the logic jules Gilbert <jules.stocks@gmail.com> - 2011-06-26 10:38 -0700

#332 — I want to replace the logic

The input is the filesize of some previously compressed client file.
This routine's is to return with a program derived suggested block-
size.

I want to this routine with something better.  Here are my
requirements:

a)  The block size should be divisible by 8 (to accomodate double word
    boundaries.)

b)  Produce a small remainder -- that portion of the file isn't
processed,
    but simply passed through.  (It does get processed by the
subsequent
    pass.)

c)  As large as possible but not excessive.  Now, the largest block
size
    returned by this routine is 256k.

d)  The smallest block size should be something like 4k (which isn't
what
    this routine does.) 4k should be the low block size limit because
that's
    roughly as small as the zLib routines will compress too.  (Sure,
they may
    very well work fine with, say, 1k buffers, but my system gives
compress
    smaller and smaller buffers.  That's what it sees, and since the
    efficiency is low when buffers are small, I'm not interested in
cutting
    up buffers, etc.)

I don't need power-of-two but I do need to keep the div-by-8.

If you want me to use your code, do not publish it with a copyright.
I can
not use copyrighted code.  (I can use the MIT license, a couple of
others.)


int
chooseBLOCKSIZE(int fs)
{
 int bs = 512 * 1024;
 int xt,nb,rm;

  xt = 129;
  while (xt > 128)
   {
     bs /= 2;
     rm = fs % bs;
     nb = fs / bs;  if (nb == 0) nb++;
     xt = rm / nb;
   }

  return bs;
}

[toc] | [next] | [standalone]

#334

Gilbert wrote:

> If you want me to use your code, do not publish it with a copyright.

Why should anyone care?

> I can not use copyrighted code.
> (I can use the MIT license, a couple of others.)

What makes you think MIT-licensed code is not copyrighted code?

The MIT license states:

Copyright (C) <year> by <copyright holders>
[...]
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.

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

#335

On Jun 24, 4:40 am, Noob <r...@127.0.0.1> wrote:
> Gilbert wrote:
> > If you want me to use your code, do not publish it with a copyright.
>
> Why should anyone care?
>
> > I can not use copyrighted code.
> > (I can use the MIT license, a couple of others.)
>
> What makes you think MIT-licensed code is not copyrighted code?
>
> The MIT license states:
>
> Copyright (C) <year> by <copyright holders>
> [...]
> The above copyright notice and this permission notice shall be
> included in all copies or substantial portions of the Software.


I think the intent of my statement was clear.  But just in case
someone is legitimately confused, the MIT license does not contain
unacceptable provisions.  The BSD family of licenses is acceptable,
too.

I'm expecting to distribute my source code, probably some of my users
will include companies who sell software.  Some licenses, such as the
GNU license are not popular with people who sell software.

If someone writes some code, I can't touch it if it's been GNU
licensed.  I would rather walk away with nothing then to see my work
product become free.

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

#336

On Jun 24, 8:04 am, jules Gilbert <jules.sto...@gmail.com> wrote:
> On Jun 24, 4:40 am, Noob <r...@127.0.0.1> wrote:

> If someone writes some code, I can't touch it if it's been GNU
> licensed.  I would rather walk away with nothing then to see my work
> product become free.

And yet you remain the fool you have always been for more than a
decade of posts.

No one really cares about the licence you use, what all can see is for
that years your wild claims have been back with *ZERO* examples of
working code.  First get something that works both to compress and
decompress, then talk.  So far you still have nothing worth anyone's
time.

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

#337

On Jun 24, 9:04 am, jules Gilbert <jules.sto...@gmail.com> wrote:
> On Jun 24, 4:40 am, Noob <r...@127.0.0.1> wrote:

Jules, if don't understand the difference between the LGPL and the
GPL, you probably shouldn't be distributing software of any kind.

- Mark

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

#338

> Jules, if don't understand the difference between the LGPL and the
> GPL, you probably shouldn't be distributing software of any kind.

To be fair, I don't think a non-lawyer can understand the difference 
between them just by reading license, it was not written with human 
beings in mind.

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

#343 — Choosing between GPL and LGPL

Captain Obvious wrote:

> To be fair, I don't think a non-lawyer can understand the difference
> between [GPL and LGPL] just by reading license, it was not written
> with human beings in mind.

Are you implying lawyers are soul-eating demons?  ;-)

For the record, the FSF does provide /some/ guidance.

http://www.gnu.org/licenses/gpl-faq.html
http://www.gnu.org/licenses/gpl-howto.html
http://www.gnu.org/philosophy/why-not-lgpl.html

Regards.

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

#344

On 6/26/2011 3:39 AM, Captain Obvious wrote:
>> Jules, if don't understand the difference between the LGPL and the
>> GPL, you probably shouldn't be distributing software of any kind.
>
> To be fair, I don't think a non-lawyer can understand the difference
> between them just by reading license, it was not written with human
> beings in mind.

yeah. not many people know the differences up front, and LGPL basically 
has nearly the same name and is often used in conjunction with GPL, 
meaning the differences are less obvious.


with MIT or BSD, it is generally more readily understood what the 
differences are.

hence, why I have mostly settled on MIT for a lot of my stuff 
(theoretically, nevermind me not yet going through the long and dull 
process of putting MIT headers everywhere).


I guess my posting here has been a bit sparse, mostly because most of my 
compression-related activities have boiled down to:
common file formats;
deflate;
data I can filter somehow and shove through deflate...

mostly all as I go and work on other things.

sadly, all this doesn't really lead to a whole lot anymore relevant to 
the topic of compression.


although, some thoughts/experimentation have involved the possibility of 
"a lossless video-codec vaguely similar to a sort of Motion-PNG with 
motion compensation and deltas".

I had experimented with something like this, and my tests did show it to 
be a bit smaller (~ 25% to 30% the size) of the collection of input PNGs 
(likely subject to vary though).

although, it was still bigger than a plain MJPEG video, or any of the 
conventional lossy codecs tested.

or such...

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

#364

On Jun 27, 6:11 pm, BGB <cr88...@hotmail.com> wrote:
> On 6/26/2011 3:39 AM, Captain Obvious wrote:
>
> >> Jules, if don't understand the difference between the LGPL and the
> >> GPL, you probably shouldn't be distributing software of any kind.
>
> > To be fair, I don't think a non-lawyer can understand the difference
> > between them just by reading license, it was not written with human
> > beings in mind.
>
> yeah. not many people know the differences up front, and LGPL basically
> has nearly the same name and is often used in conjunction with GPL,
> meaning the differences are less obvious.
>

For a certain reason the Argtable issue has become a non problem.
People to thank, etc.

So I took one of my methods that compress RAD and read a previously
compressed file, block it into buffers, compress the buffers, and
*now* emit a single memory vector, in fact a byte-wide vector.  Which
becomes the target of the subsequent pass.

Now this approach has problems, two problems particularly, first the
method I am using must be a power of two, and not just any po2, ie.,
other restrictions apply.  But this impacts the remaining fragment
which must be part of the constructed memory vector.

(I said two problems, lets see what happens with this post, maybe I'll
get into the other issue.)

But it does take a large file, crunch it, and produce a file below 8k
bytes.

Since everyone always asks "Can you reconstruct?" I'll say (I think
it's obvious from the above,) that no, I don't have a completed
program that runs and re-processes the memory vector.  Although I have
written the decompressor for this method and it works, so far this is
only reconstructing from a malloc'ed block.

(And let me tell you -- when you first see those pairs of values,
well, I think people here know what I'm talking about.)

And if you're new here what I'm saying is this:  All files, every
single file, including packages of files (such as a .tar file
contains,) can be represented in 8k bytes.

And yes, though I've never done it, I'm pretty sure I can compress and
reconstruct more than 2 ^ 64k files.

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

#365

On 10 Jul., 23:22, jules Gilbert wrote:
> [...]
> I don't have a completed program that runs
> [...]
> I'm saying [...]:  All files, every
> single file, including packages of files (such as a .tar file
> contains,) can be represented in 8k bytes.

Thank you for disqualifying yourself once again so obviously.
I always suspected that you couldn't count. This proves that you
really can't.

> And yes, though I've never done it, I'm pretty sure I can compress and
> reconstruct more than 2 ^ 64k files.

(for simplicity, let's assume byte=octet)

I think you meant 256^64k.
Otherwise, this would not be impressive at all since 2^64k = 256^8k.

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

#366

jules Gilbert wrote:
 
<snip>

> compressed file, block it into buffers, compress the buffers, and
> *now* emit a single memory vector, in fact a byte-wide vector.  Which
> becomes the target of the subsequent pass.

Out of curiosity, are you a native English speaker?

What exactly do you mean by "single memory vector"? Those are all
valid words individually, but put them together into a phrase and you
get meaningless jargon. Are you talking about an undetermined length
of contiguous bytes stored in memory? Possibly an array to use
terminology from many programming languages and a well known term?

> Now this approach has problems,

For those who insist Jules doesn't know what he's talking about, I
offer the above as counterpoint. 

<snipping story of Jules mythological magic recursive machine>

> But it does take a large file, crunch it, and produce a file below 8k
> bytes.
>
> Since everyone always asks "Can you reconstruct?"

You find this question surprising?

> I'll say (I think it's obvious from the above,) that no, I don't have
> a completed program that runs and re-processes the memory vector.

Let me get this straight. Here you say no I can't reverse the process.

> Although I have written the decompressor for this method and it
> works, so far this is only reconstructing from a malloc'ed block.

And here you say you can "decompress" stuff stored in memory.

Where exactly does the "single memory vector" exist?

> (And let me tell you -- when you first see those pairs of values,
> well, I think people here know what I'm talking about.)

If the betting windows are open, I'd like to put buttloads on 
"nobody knows what you are talking about"

> And if you're new here what I'm saying is this:  All files, every
> single file, including packages of files (such as a .tar file
> contains,) can be represented in 8k bytes.
>
> And yes, though I've never done it, I'm pretty sure I can compress and
> reconstruct more than 2 ^ 64k files.

So you've repeated; very, very often. Do you still claim this doesn't
violate the counting argument?

If I give you a single file of bytes, how many different files will
your program generate?

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

#370

On Jul 10, 4:22 pm, jules Gilbert <jules.sto...@gmail.com> wrote:
> Since everyone always asks "Can you reconstruct?" I'll say (I think
> it's obvious from the above,) that no, I don't have a completed
> program that runs and re-processes the memory vector.  Although I have
> written the decompressor for this method and it works, so far this is
> only reconstructing from a malloc'ed block.

Okay, which is it?  Is it "no, I don't have a completed program", or
"I have written the decompressor for this method and it works"?  Have
you successfully decompressed the data or not?

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

#340

On Jun 25, 2:44 pm, Mark Nelson <snorkel...@gmail.com> wrote:
> On Jun 24, 9:04 am, jules Gilbert <jules.sto...@gmail.com> wrote:
>
> > On Jun 24, 4:40 am, Noob <r...@127.0.0.1> wrote:
>
> Jules, if don't understand the difference between the LGPL and the
> GPL, you probably shouldn't be distributing software of any kind.
>
> - Mark

Oh I know the difference between a program I (or others,) code and the
libraries referenced by the program.  That's not hard to get right.
But read the LGPL, to make sure I get the right I decided to use a law
firm that specializes in copyright law.  So for now all fancy
interfaces are on hold.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.compression

csiph-web