I think Compress and UnCompress classes should be used as in following example
---------------------------------
import std.file;
import std.zlib;

void main()
{
	Compress c = new Compress;
	UnCompress u = new UnCompress;
	void[] q,w,e,wa,et; int i;

	q = read("in.dat");

	for(i=0;i+512<q.length;i+=512)
		{
		w=c.compress(q[i..i+512]);
		wa~=w; et=""; et~=w;

		w=c.flush(Z_SYNC_FLUSH);
		wa~=w; et~=w;

		e~=u.uncompress(et);
		}
	
	w=c.compress(q[i..q.length]);
	wa~=w; et=""; et~=w;
    	
	w=c.flush();
	wa~=w; et~=w;
	e~=u.uncompress(et);
	e~=u.flush();

	write("out_c.dat",wa);		
	write("out_u.dat",e);
	e = uncompress(wa);
	write("out_ux.dat",e);	

}
---------------------------------
But it does not work. Look at the .diff. With these changes it works properly. (dmd v0.133 win32)
---------------------------------
--- zlib.d	Sat Sep 24 17:25:30 2005
+++ zlib_new.d	Tue Oct  4 01:28:34 2005
@@ -287,8 +287,9 @@
 	{   delete destbuf;
 	    error(err);
 	}
-	destbuf.length = zs.total_out;
-	return destbuf;
+	//destbuf.length = zs.total_out;
+	destbuf.length = destbuf.length - zs.avail_out;
+    return destbuf;
     }

     void[] flush()
@@ -303,16 +304,22 @@
     }
     body
     {
-	void[] destbuf;
+	void[] destbuf, tmpbuf;
 	int err;

 	if (!inited)
 	    return null;

+	/*  //old code
 	destbuf = new void[zs.avail_in];
 	zs.next_out = cast(ubyte*) destbuf;
 	zs.avail_out = destbuf.length;
+	*/
+    tmpbuf = new void[zs.avail_out]; // may be  zs.avail_out+<some constnt>
+	zs.next_out = cast(ubyte*) tmpbuf;
+	zs.avail_out = tmpbuf.length;

+	/* //old code
 	err = deflate(&zs, mode);
 	if (err != Z_STREAM_END)
 	{
@@ -322,6 +329,27 @@
 	    error(err);
 	}
 	destbuf = cast(void[])((cast(ubyte *)destbuf)[0 .. zs.next_out - cast(ubyte*)destbuf]);
+	*/
+	while( (err = deflate(&zs, mode)) != Z_STREAM_END)
+		{
+		if(err == Z_OK)
+			{
+			if(zs.avail_out != 0 && mode != Z_FINISH)
+				break;
+			else if(zs.avail_out == 0)
+				{
+				destbuf ~= tmpbuf;
+				zs.next_out = cast(ubyte*) tmpbuf;
+				zs.avail_out = tmpbuf.length;
+				continue;
+				}
+			err = Z_BUF_ERROR;
+			}
+		delete destbuf;
+		error(err);
+		}
+	destbuf ~= tmpbuf[0..(tmpbuf.length-zs.avail_out)];
+	
 	if (mode == Z_FINISH)
 	{
 	    err = deflateEnd(&zs);
@@ -410,7 +438,8 @@
 	{   delete destbuf;
 	    error(err);
 	}
-	destbuf.length = zs.total_out;
+	//destbuf.length = zs.total_out;
+	destbuf.length = destbuf.length - zs.avail_out;
 	return destbuf;
     }


-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

On Tue, 04 Oct 2005 02:34:51 +0400, Artem Rebrov wrote:

> I think Compress and UnCompress classes should be used ...

Why are they even classes? Are not class meant to be representation of things rather than activities? What is a Compress? What is an UnCompress? These are processes, actions, work, and not objects. The closest thing would be that 'Compress' is a Storage mechanism for compressed data and that it can do (at least) two activities: compress data into itself and decompress data that it already has.

-- 
Derek
(skype: derek.j.parnell)
Melbourne, Australia
4/10/2005 9:10:14 AM

On Tue, 04 Oct 2005 03:14:46 +0400, Derek Parnell <derek@psych.ward> wrote:

> On Tue, 04 Oct 2005 02:34:51 +0400, Artem Rebrov wrote:
>
>> I think Compress and UnCompress classes should be used ...
>
> Why are they even classes? Are not class meant to be representation of
> things rather than activities? What is a Compress? What is an UnCompress?
> These are processes, actions, work, and not objects. The closest thing
> would be that 'Compress' is a Storage mechanism for compressed data and
> that it can do (at least) two activities: compress data into itself and
> decompress data that it already has.
>

Just look at dmd\src\phobos\std\zlib.d in the DMD distribution


-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

On Tue, 04 Oct 2005 22:01:36 +0400, Artem Rebrov <ar_other@mail.ru> wrote:

> On Tue, 04 Oct 2005 03:14:46 +0400, Derek Parnell <derek@psych.ward> wrote:
>
>> On Tue, 04 Oct 2005 02:34:51 +0400, Artem Rebrov wrote:
>>
>>> I think Compress and UnCompress classes should be used ...
>>
>> Why are they even classes? Are not class meant to be representation of
>> things rather than activities? What is a Compress? What is an UnCompress?
>> These are processes, actions, work, and not objects. The closest thing
>> would be that 'Compress' is a Storage mechanism for compressed data and
>> that it can do (at least) two activities: compress data into itself and
>> decompress data that it already has.
>>
>
> Just look at dmd\src\phobos\std\zlib.d in the DMD distribution
>
>
(Un)Compress classes holds state of (un)compressor.
When I can't compress whole data I split it to chunks. I transmit these
chunks to other side of the connection (for example) and uncompress
them one by one. The goal is to be able to decompress the first chunk
before I can receive the last one.



-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

In article <op.sx4xxhtpncj208@comp>, Artem Rebrov says...
>
>(Un)Compress classes holds state of (un)compressor.
>When I can't compress whole data I split it to chunks. I transmit these
>chunks to other side of the connection (for example) and uncompress
>them one by one. The goal is to be able to decompress the first chunk
>before I can receive the last one.

For what it's worth, I typically implement this functionality as a stream filter in C++.  It would probably be worth doing something similar in D.


Sean