Here a few other links ... after reading through the RFC. Not sure if
they help, but just sharing from my own research into assisting on
this topic
https://github.com/evegard/pngview/blob/master/huffman.c#L54
And a fork of the putty version with dynamic huffman coding:
http://rc.quest.com/viewvc/
Ah, good point.
Here's another thing:
+/ bits < #@> F hcodes A
31
In some cases the deflate codes have fewer bits than huffman coding.
So the problem is finding the algorithm that was used to define bits - it's
not the huffman coding algorithm.
I guess I need to spend some time reading rfc
It refered to A and F (not A1 and F1) in my first email.
I sent you an email off-list that included attachments.
Ср, 10 сен 2014, Raul Miller написал(а):
> Something is wrong here:
>
>$bits
> 286
>(0=bits)-:0=F1
> 0
>+/0=F1
> 9
>+/0=bits
> 26
>
> 0 bits should mean that that sy
Something is wrong here:
$bits
286
(0=bits)-:0=F1
0
+/0=F1
9
+/0=bits
26
0 bits should mean that that symbol is not represented.
But there are only 9 such symbols in F1.
--
Raul
On Wed, Sep 10, 2014 at 11:09 PM, bill lam wrote:
> I guess the description of the algorithm itself is
I guess the description of the algorithm itself is correct, but
it does not mention how to get the bits of each symbol. The bit
lengths for classic and zlib huffman can be different.
for testing, the bits for A from libz.so are
":&>_10<\bits
10 10 10 10 10 10 10 10 10 10
10 10 10 10 10 10 10 1
I agree that there should not be any 2s.
However, I think this means that there's something wrong with the
description of the algorithm, or at least my understanding of it.
I'll need some time to digest this.
Thanks,
--
Raul
On Wed, Sep 10, 2014 at 9:39 PM, bill lam wrote:
> pardon me of for
pardon me of forgetting telling another cavaet.
zlib huffman code is suboptimal so that bit length of code
for each symbol can be longer than that in classic huffman.
eg in my orignal data.
10{. ":@>F deflatecodes2 A
1 1 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 0
1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1
Oops, sorry about that.
Try it this way:
deflatecodes2=:4 :0
L=. #@> x hcodes y
U=. 0,~.L
R=. ;<@(({. >./@(>#])&U #1:)@{. <@:+"1-@{.{."1 #:@i.@#)/.~L
R/:;(F1 deflatecodes2 A1
0
1 1 0
1 1 1 1 1 1 0
1 1 1 1 1 1 1
1 1 1 0 0 0
1 1 1 0 0 1
1 1 1 0 1 0
1 1 1 0 1 1
1 1 1 1 0 0
1 1 1 1 0 1
1 1 1
I get different results in jhs where none of the lines have an empty line.
NB. Here
5#''
(5#'')*/i.5
''*/i.5
''
NB. To here
I actually always expected to have an empty line for each expression.
Linda
-Original Message-
From: programming-boun...@forums.jsoftware.com
[
On 10/09/2014 4:43 PM, Don Kelly wrote:
I get
5#''
(5#'')*/i.5
''*/i.5
Apologies for the double spacing in the original reply. hope that this
is better.
Don Kelly
--
For information about J forums see http://www.jsoftware.
I get
5#''
(5#'')*/i.5
''*/i.5
NOTE-. blank line
using J802
Don Kelly
On 10/09/2014 4:16 PM, Linda Alvord wrote:
Dan, after your 4 series, I tried this. Also odd:
NB. Begin here
5#''
(5#'')*/i.5
''*/i.5
''
NB. Note the spacing betwe
Thanks Joe,
putty only use zlib static huffman for encoding so that it does not build
any huffman dictionary table.
The zlib static huffman code does not care about individual symbol's
frequency, it just encode 0 to 286 into bits, see section 3.2.6.
On Sep 11, 2014 1:26 AM, "Joe Bogner" wrote:
What were you expecting?
Please excuse typos; sent from a phone.
> On Sep 10, 2014, at 7:16 PM, "Linda Alvord" wrote:
>
> Dan, after your 4 series, I tried this. Also odd:
>
> NB. Begin here
> 5#''
>
> (5#'')*/i.5
> ''*/i.5
> ''
>
> NB. Note the spacing between lines above
>
>
sorry typo,
because there are _7_ symbols in the 6 bit code.
On Sep 11, 2014 7:26 AM, "bill lam" wrote:
> for zlib, all huffman code of the same bit length are in sequential,
> starting with a code an 0 bit in the end (even number), and no gaps in the
> block (consecutive). the symbols of these
for zlib, all huffman code of the same bit length are in sequential,
starting with a code an 0 bit in the end (even number), and no gaps in the
block (consecutive). the symbols of these block represented are sequential
but not consecutive (possibly gaps). So the zlib huffman code is slightly
less e
Dan, after your 4 series, I tried this. Also odd:
NB. Begin here
5#''
(5#'')*/i.5
''*/i.5
''
NB. Note the spacing between lines above
Linda
-Original Message-
From: programming-boun...@forums.jsoftware.com
[mailto:programming-boun...@forums.jsoftware.com] On Behalf
It seems like chat is fine to share things created with J (as I've done
that before myself). Ideally we'd have the author or some other volunteer
curate meaningful content into the wiki. If there's at least some positive
response on chat then it would seem to indicate it's worthy of putting it
on t
I do not see a conflict here.
Thanks,
--
Raul
On Wed, Sep 10, 2014 at 1:41 PM, Dan Bron wrote:
> I'd be up for that. But probably a better medium is the Wiki, as art is
> typically an iterative process?
>
> -Dan
>
> - Original Message ---
>
> Subject: Re: [Jprogramming] Repla
I'd be up for that. But probably a better medium is the Wiki, as art is
typically an iterative process?
-Dan
- Original Message ---
Subject: Re: [Jprogramming] Replace one item of a list
From: Raul Miller
Date: Wed, 10 Sep 2014 13:27:51 -0400
To: Programming forum
Product managers, in general, do not make their decisions on technical
grounds, but on popularity grounds.
That means we need to support people generating media - art (images
and image manipulating tools), games (games themselves but also tools
for working with the data used in games), music (soun
You've already likely considered this, but if it were me I would compare
results to a working implementation. The one from putty seems pretty clean
and standalone:
https://raw.githubusercontent.com/grumpydev/PortablePuTTY/master/SSHZLIB.C
. I was able to compile it on windows no problem and I assum
I do not think the dictionary is wrong but perhaps it is incomplete. As you
pointed out, the tally is 1 and the neutral element intance provision does
not apply; so, for example, in the sentence (*/0) or in general (u/A) where
A is an atom the result is the atom (A) (even if u is undefined!). Where
I think the use of the term "consecutive" rather than "sequential" is telling.
The described algorithm is: compute the huffman code lengths:
#@>F1 hcodes A1
1 3 7 7 6 6 6 6 6 6 6 2
Then assign ascending huffman codes first in length order and then
within codes of the same length.
Taken litera
Product managers, in general, aren't going to base their decisions on what
some random internet stranger wrote on an obscure mailing list. More
likely to influence their decision is the very fact that the primary
mailing list for the language is obscure.
In other words: if our goal is to get peopl
Sorry, Dan -- none of that would work for me if I were a product manager.
One with some control over which language to choose and to recruit my
programmers for. Or more like it: which language to avoid. One sight of
this thread and I'd say "toy language -- my staff will be spending all
their time f
Linda wrote:
> *i.5
> 0 1 1 1 1
>*/"0 i.5
> 0 1 2 3 4
Joe wrote:
> It looks like /"0 yields the right side without evaluating the left side
>
> asdfasdf/"0 i.5
> 0 1 2 3 4
>
> I don't totally get it myself
This boils down to foo/ scalar_value .
Note that because of the "0 (meani
Great explanation Jan-Pieter. It makes perfect sense. Thank you
On Wed, Sep 10, 2014 at 11:42 AM, Jan-Pieter Jacobs <
janpieter.jac...@gmail.com> wrote:
> I think the manual page on insert (
> http://www.jsoftware.com/help/dictionary/d420.htm ) sufficiently
> explains this behavior:
>
> "0 makes
I think the manual page on insert (
http://www.jsoftware.com/help/dictionary/d420.htm ) sufficiently
explains this behavior:
"0 makes the function operate on scalars and, as per the Vocabulary on
Insert (/) :
" If y has no items (that is, 0=#y), the result of u/y is the neutral
or identity elemen
When I see a newcomer (not you, Ian, obviously) raise such concerns as "how
do I do updates in place?", my typical response to this is:
J is a function-level language. It expresses computations in terms of data
flow. Verbs (functions) are primary; nouns (data) are secondary.
STOP THINKING IN TE
On Wed, Sep 10, 2014 at 9:57 AM, Linda Alvord
wrote:
>
> Here's the example that is puzzling me.
>
> *i.5
> 0 1 1 1 1
>*/"0 i.5
> 0 1 2 3 4
>
>
It looks like /"0 yields the right side without evaluating the left side
asdfasdf/"0 i.5
0 1 2 3 4
This might be helpful in understanding:
h
>> Then it actually does in-place updating (even though, at face value, J
syntax does not permit such a thing).
> Please explain why this is not permitted by J syntax.
My use of the word "permit" was misleading -- "suggest' might have been
better. Roger is right to challenge me on that point.
Wh
Bo, would you say that ordinal fractions are equivalent to "left lists"
from the paper "From Trees Into Boxes"? If so, then J already addresses the
nodes of a tree in that manner through {::, with the slight modification
that data in a node can be an array of arbitrary shape so each step down
the t
For huffman coding used in zlib:
https://www.ietf.org/rfc/rfc1951.txt section 3.2.2.
The Huffman codes used for each alphabet in the "deflate"
format have two additional rules:
* All codes of a given bit length have lexicographically
consecutive values, in the same order as the symbols
th
I would be interested to see a model of
arrays-and-trees-as-ordinal-fractions in J. With such things as indexing,
catenation, etc.
-Dan
- Original Message ---
Subject: Re: [Jprogramming] Ragged Array Shapes are Trees
From: "'Bo Jacoby' via Programming"
Date: Tue, 9 Sep 20
Roger, To explain f would you call } item amend rather than amend ?
]a=: 100 (2 4 5}) a=:i.10
0 1 100 3 100 100 6 7 8 9
f=: 13 :'x (2 4 5}) y'
100 f i.10
0 1 100 3 100 100 6 7 8 9
Here's the example that is puzzling me.
*i.5
0 1 1 1 1
*/"0 i.5
0 1 2 3 4
Linda
-
35 matches
Mail list logo
