New Patch Changes!

Joolz · #1 01-08-2003, 03:20 AM

From the nice guys at ShowEQ:

Quote:

Some pretty major changes on test. Posting this mostly to help cut down on duplicated work. I havent worked through too much of it yet, but heres what I do know so far. Most all data is zlib compressed now, which is a nice bandwidth saver. The compression is basicly implimented so that multiple packets can be rolled together into a single compression block before being sent.

New opcode: 0x7642 - compressed data packet. Seems to have a 3-5 byte header then the zlib compressed data. If the 3rd byte is 0xff, its a 5 byte header, otherwise 3, it seems. Unsure of what the other stuff indicates so far. In the decompressed data, you've got 1 or more packets strung together, havent worked out exactly how this works, but you do still have the opcode and data for each packet.

Aside from all that, there appear to be some data structure changes, mostly in the form of 16 bit values being changed to 32 bits instead.

In short, lots of work to do

What does this mean for us? Well, we HAVE to implement the 16 bit to 32 bit transitions. That shouldn't take us too long.

As for the new opcode... well... these appear to be packets from the server. Also, the net code in its current form really doesn't need the overhead of zlib compression right now!

devn00b · #2 01-08-2003, 07:03 AM

Compressing every packet will/would be nice tho. would mean more bandwith for smaller server, for all those newly moving NPC's (since it has to send a npc update packet right?) or for other servers with lots of PC's moving about.

overall how much does compressing each packet save? my guess is that its not that much. but hey every little byte helps.

kathgar · #3 01-08-2003, 01:22 PM

But does the gain rom compression outweigh the expense of compressing is the problem. Hell, we may just get lucky and the client accepts the old way

Joolz · #4 01-09-2003, 07:46 AM

And... here we go!

Quote:

[19:10] <Joolz> Unknown opcode: 0x7642 size:129
[19:10] <Joolz> 0: 58 18 C4 78 5E CB 29 56 - 36 34 32 36 33 67 70 B3 | X..x^.)V64263gp.
[19:10] <Joolz> 16: CC CF 2C 88 4A CA 30 30 - 48 F6 09 29 B6 60 20 03 | ..,.J.00H..).` .
[19:10] <Joolz> 32: D8 01 F5 7C F9 2B CC 50 - C7 D0 C0 C0 F0 E6 6F BD | ...|.+.P......o.
[19:10] <Joolz> 48: 14 13 03 83 C7 4F 61 86 - 05 0A A2 0C 3F 7E 08 83 | .....Oa.....?~..
[19:10] <Joolz> 64: 4D CC 16 FD 5A EE CC C0 - 70 60 96 E0 D7 F2 F9 40 | M...Z...p`.....@
[19:10] <Joolz> 80: 8C 2C D6 F2 5B 98 E1 C3 - 2F 61 86 02 A0 19 AC 13 | .,..[.../a......
[19:10] <Joolz> 96: BE 97 AF B8 FA 0D 2C AF - 00 C4 95 4B 9E 97 83 68 | ......,....K...h
[19:10] <Joolz> 112: 90 FE DB E2 CF C1 66 80 - E4 0E 01 31 00 43 4F 2D | ......f....1.CO-
[19:10] <Joolz> 128: B8 | .

So, ladies.... compressed... HOW? Its bugging me now... I can't get zlib to do 'jack':

Client::Process()

Code:

		case 0x7642:{

			cout &lt;&lt; "Compressed packet data\n";

			z_stream t;
			APPLAYER *tmper;
			tmper = new APPLAYER(0, 1024);

			t.next_in = app->pBuffer;
			t.avail_in = app->size;

			t.next_out = tmper->pBuffer;
			t.avail_out = tmper->size;
			t.zalloc = (alloc_func)0;
			t.zfree = (free_func)0;
			
			inflateInit(&amp;t);
			inflate(&amp;t, Z_FINISH);

			cout &lt;&lt; t.total_out &lt;&lt; '\n';

			break;

always comes back with 0... arse!

kathgar · #5 01-09-2003, 08:50 AM

they might have compressed each seperatly .. or just fucked with the headers, or just not used zlib..

Joolz · #6 01-09-2003, 09:00 AM

They did use zlib... If you decompress the above packet, you get:

Quote:

0: 6C 73 23 31 32 33 36 37 - 00 72 4E 54 65 62 75 39 | ls#12367.rNTebu9
16: 6A 74 4B 65 6D 4A 48 53 - 00 00 00 00 00 00 00 00 | jtKemJHS........
32: 00 00 00 00 00 00 00 00 - 00 00 00 00 00 00 00 00 | ................
48: 00 00 00 00 00 00 00 00 - 00 00 00 00 00 00 00 00 | ................
64: 00 00 00 00 00 00 00 00 - 00 00 00 00 3E 00 00 00 | ............>...
80: F4 FD 13 00 7E 00 80 00 - 00 EC FD 7F 1A 02 00 00 | ....~...........
96: 48 F9 13 00 A0 20 15 00 - F8 F8 13 00 00 00 00 00 | H.... ..........
112: 6B 15 F5 77 43 00 00 C0 - 9A 11 F5 77 9F 11 F5 77 | k..wC......w...w
128: 00 00 00 00 6B 15 F5 77 - 43 00 00 C0 84 FB 13 00 | ....k..wC.......
144: F0 FA 13 00 70 FD 13 00 - 05 90 F7 77 A8 D5 F6 77 | ....p......w...w
160: 00 00 00 00 20 00 00 00 - 79 A4 E7 77 20 00 00 00 | .... ...y..w ...
176: 6B 15 F5 77 DB 17 E7 77 - 43 00 00 C0 00 00 00 00 | k..w...wC.......
192: C2 00 00 00 | ....

So... what does that mean? There's no recognisable opcodes... (this is the first packet sent by the client to the server)

Joolz · #7 01-09-2003, 09:09 PM

As correctly pointed out, its a login packet... but, where is the opcode header etc? Good question!

Joolz · #8 01-09-2003, 09:19 PM

Reality check! Done it.... I now have logging in working on the new EQEMU... I will check changes in tonight when I get home from work.

Shame no one really helped me... But at least I got it done.

Shawn319 · #9 01-10-2003, 02:04 AM

Quote:

Originally Posted by Joolz

Reality check! Done it.... I now have logging in working on the new EQEMU... I will check changes in tonight when I get home from work.

Shame no one really helped me... But at least I got it done.

You work too fast lol!!! Theres no time for anyone to help you

#10 01-10-2003, 02:08 AM

do you suspect a new emu for these changes or perhaps just a patch?

_____
EL/FA

Joolz · #11 01-10-2003, 06:01 AM

I'm porting the structure changes now... This will be a new emulator.. + when the new emulator comes out... the old version of EQ will NOT NOT NOT work with it.

mByte · #12 01-10-2003, 11:09 AM

Quote:

Originally Posted by Joolz

I'm porting the structure changes now... This will be a new emulator.. + when the new emulator comes out... the old version of EQ will NOT NOT NOT work with it.

All previous work is back to stage 1?

just_add_water · #13 01-10-2003, 12:48 PM

.. is it that hard to understand? i mean gezz

Quote:

This will be a new emulator.. + when the new emulator comes out... the old version of EQ will NOT NOT NOT work with it.

That mean your gonna need to patch to play on 0.4.2 servers (when 0.4.2 comes out and ONLY THEN SO *DON'T* PATCH RIGHT NOW)

Joolz · #14 01-10-2003, 07:04 PM

I did think that what I said was simple to understand. Although do understand that we do need to start the structs from scratch :(

killspree · #15 01-11-2003, 02:48 PM

I've had the opprotunity to test out the new compression on EQlive, and I have to say it's simply amazing. It looks like it's very worth it, as the connection lag I had with ~80 people in an area while a raid was going on, was about the same as the lag I'd have with about 30 people prior to this patch.