Sending long messages in UTF-8 causes IRc to generate non RFC-compliant packet

Miranda Version                  : 0.8.10
Unicode Build                    : Yes
Test Build # (if applicable)     : No
Plugin Version # (if applicable) : IRC.dll v.0.8.3.0 [17 Nov 2009] - IRC 
protocol |Unicode aware|

What steps will reproduce the problem?
1. Turn on UTF-8 support for servers (in Extra card).
2. Write a message that is nearly 510 characters long using national 
characters (with diacritics or other non-ASCII characters).
3. Send the message.

What is the expected result?

Since messages longer than 510 ASCII-only characters (like a long line of 
a's) are split into pieces and sent correctly, the expected result would 
be the exact same behavior, only UTF-8 aware.

What happens instead?

Depends on the server:
1. ngIRcd acknowledges the long packet and disconnects the client.
ngIRcd - ngircd[5392]: Request too long (connection 7): 528 bytes (max. 
512 expected)!

2. freenode.net trims the message to fit into first 510B.

VersionInfo:

CPU: Intel(R) Core(TM)2 Quad  CPU   Q9300  @ 2.50GHz [Intel64 Family 6 
Model 23 Stepping 7] [DEP enabled] [4 CPUs]
Installed RAM: 8191 MBytes
Operating System: Windows (TM) Vista Business [version: 6.0.6002 Service 
Pack 2]
Shell: explorer.exe, Internet Explorer 8.0.6001
Administrator privileges: Yes
OS Languages: (UI | Locale (User/System)) : Czech/Czech | Czech/Czech 
[Czech, Tajik (Cyrillic), Persian, Vietnamese, Armenian, Azeri (Latin), 
Basque, Upper Sorbian, Macedonian (FYROM), Mongolian (Cyrillic), Tibetan, 
Welsh, Khmer, Galician, Filipino, Divehi, Dari, Arabic, Bulgarian, 
Catalan, Chinese (Simplified)]
Free disk space on Miranda partition: 116213 MBytes
Miranda path: S:\Program Files\MirandaIRc
Miranda IM version: 0.8.10 Unicode [running inside WOW64]
Build time: 17:01:31 on 17 November 2009
Profile path: S:\Program Files\MirandaIRc\Profile
Profile size: 640.00 KBytes
Profile creation date: 17 Sep 2009 at 13:04:55 (UTC +1:00)
Language pack: Czech (CZ) [0405] v. 2.18.1.1, modified: 02 May 2008 at 
13:55:52 (UTC +1:00)
Nightly: No
Unicode core: Yes

Active Plugins (11):
¤ clist_modern.dll v.0.8.1.1 [23 Jun 2009] - Modern Contact List '08 |
Unicode aware|
¤ db_autobackups.dll v.0.0.0.4 [02 Aug 2008] - DB AutoBackups
¤ dbx_mmap_sa.dll v.0.7.5.3 [18 Jun 2008] - Miranda SecuredMMAP + 
Autobackups database driver |Unicode aware|
¤ IRC.dll v.0.8.3.0 [17 Nov 2009] - IRC protocol |Unicode aware|
¤ KeepStatus.dll v.0.0.2.92 [26 Jan 2008] - KeepStatus
¤ keyboardnotify.dll v.1.5.7.7 [13 Feb 2008] - Keyboard Notify Ext.
¤ StartupStatus.dll v.0.0.3.36 [26 Jul 2008] - StartupStatus
¤ svc_vi.dll v.1.4.3.3 [14 Apr 2008] - Version Information |Unicode aware|
¤ tabSRMM.dll v.2.2.1.17 [22 Jun 2009] - TabSRMM |Unicode aware|
¤ updater.dll v.0.5.3.0 [29 Jul 2007] - Updater (Unicode)
¤ watrack.dll v.0.0.6.10 [28 Aug 2008] - Winamp Track |Unicode aware|

Other Resources:
WireShark captured packet:
Transmission Control Protocol, Src Port: 51386 (51386), Dst Port: 6667 
(6667), Seq: 62, Ack: 1815, Len: 525
Message was:
"Metro D se proti očekávání nebude stavět z centra na okraj, ale naopak. 
První úsek by měl vést z Nových Dvorů, přes Nemocnici Krč, Nádraží 
Krč a 
Olbrachtovu na stanici metra Pankrác, kde by cestující přestoupili na již 
vybudované metro C. Podle Šteinera by metro D nemuselo být tak hluboko pod 
povrchem, a proto by stavba nemusela být vždy tak náročná na ražbu. V 
dalších etapách by se pak připojila Písnice a nakonec i Pankrác s 
Náměstím 
Míru a Hlavním nádražím."
Original issue reported on code.google.com by GroundZe...@gmail.com on 11 Dec 2009 at 4:24
ValentijnNK / miranda

Sending long messages in UTF-8 causes IRc to generate non RFC-compliant packet #450
