The Arsenal Files 4

home *** CD-ROM | disk | FTP | other *** search

/ The Arsenal Files 4 / The Arsenal Files 4 (Arsenal Computer).ISO / net_info / fsc-0081.l01 / FSC-0081.01A < prev next >

Wrap

Text File | 1995-03-01 | 34KB | 812 lines

| Document: FSC-0081 Part A | Version: 001 | Date: 01 Mar 1995 | | Mikael Stσldal, 2:201/337 A TYPE-3 Packet proposal Mikael Stσldal, 2:201/337@FidoNet Status of this document ======================= Copyright (C) 1992-1994 by Mikael Stσldal. This document may be freely distributed and copied. The standard described may be implemented by anybody. You may not distribute modified versions of this document without written permission from the author. Fido and FidoNet are registered trademarks of Tom Jennings and Fido Software. Introduction ============ This is a proposal on a new packet format for use in FidoNet and other FTNs (FTN = FidoNet Technology Network). This format can be used for both NetMail (private point-to-point mail) and EchoMail (electronic conferences). This document defines the transport layer. It also includes a presentation layer definition, but that can be changed without affecting the transport layer. It also includes a application layer (stored messages) definition, but it's not mandatory. This packet format conforms with FSC-39, rev 4. It has the PktType field in the header at the same position as TYPE-2 and it contains a capability word at the same position as TYPE-2+ (but no CW validation copy). The packet header has also the same size as in TYPE-2. This should make it easy to implement mail-processors supporting both TYPE-2 and TYPE-3. This packet format separates the header from the message data and the header is extensible. A mail processor (the transport layer) doesn't have to look at the message data. This format eliminates most kludges currently used in TYPE-2. This packet format allows multiple AreaTags in EchoMail. This packet format is much more extensible than TYPE-2 and is probably easier to implement. This packet format allows up to 4 GigaBytes large messages and allows binary data in messages, this makes it possible to implement multimedia capabilities in the future. Definitions =========== Characters ---------- In this document, literal characters are written as two uppercase hexadecimal (base 16, 0..9A..F) digits following by lowercase h, e.g. "1Fh". The following abbreviations are used: "NUL" means 00h, "CR" means 0Dh and "space" means 20h (the quotation marks are not included). Basic data types ---------------- Name Length Description ---------------------------------------------------------------------------- ShortInt 1 8-bit unsigned integer. Integer 2 16-bit unsigned integer, Intel 8086 byte order. LongInt 4 32-bit unsigned integer, Intel 8086 byte order. Byte 1 Field of 8 bits. Word 2 Field of 16 bits, Intel 8086 byte order. String[n] n Fixed length string of n bytes, NUL-padded. String{n} max n Variable length string of max n bytes, NUL-terminated. n includes the terminating NUL. String Variable length string, not terminated. Address ------- Field Type Description Range ---------------------------------------------------------------------------- Zone Integer Zone address. 1..65535 Net Integer Net address. 1..65535 Node Integer Node address. 0..65535 Point Integer Point address or zero if boss. 0..65535 FTN address in text format -------------------------- An FTN address in text format must, unless otherwise stated, be in the Zone:Net/Node[.Point] format, where .Point must be excluded if point is zero, included otherwise. TimeStamp --------- A LongInt which contains the number of seconds since 00:00:00 1st January 1970 UTC. Organization ------------ An organization is a network such as FidoNet, VirNet, SigNet or InterNet. It's sometimes called "domain", but I use the term organization to avoid confusion with InterNet domains. Reserved -------- All fields marked "reserved for future extension" must be zeroed when creating/writing and ignored when reading. They might be defined by future revisions of this document. Filenames ========= It's recommended to use the naming scheme in FTS-LIST, for both NetMail and EchoMail. There should be no problem with using the same name space for TYPE-2 and TYPE-3 packets because the packet headers are enough compatible. Packet format ============= Packet header Zero or any number of data records. 00h 00h The first two bytes of a data record is used to check for the end of a packet. If both are zero, the end is reached (and the rest, if any, should be ignored). Packet header ============= Field Type Description ---------------------------------------------------------------------------- PktOrig Address The node who created this packet. PktDest Address The node who should receive this packet. SubType Integer Packet contents (see below). PktType Integer Always 3. PktDate TimeStamp When the packet was created. ProdCode Integer FTSC Product Code. Programs without Product Code must use 65535. MajorVer ShortInt Major product version. MinorVer ShortInt Minor product version. Org String[16] Organization (see below). CapWord Word Capability Word (see FSC-39). Password String[8] Packet password. ExtraInfo String[4] Reserved for future extension. SubType ------- Defines what the packet contains. If this field is zero, data record is defined by the "Packed message" part of this document. If this field is non-zero, data record is not defined by this document. This field can be used for two things. The first is to define new packet types without having to use a new packet header. The second is to allow distribution of other things than Email. Organization ------------ This field contains the name of the organization the packet travels in. This field applies to both PktOrig and PktDest since packets are intra-organization. Packed message ============== The messages should be in chronological order in a packet. Field Type Description ---------------------------------------------------------------------------- HeadSize Integer Size of the message header (see below). MsgFlags Word Message flags (see below). MsgDate TimeStamp When the message was created. MsgID LongInt Unique message identifier (see below). ReplyID LongInt Reply linkage information (see below). MsgLength LongInt Size of MsgData. MsgOrig Address The node who created this message. MsgDest Address The node who should receive this message. CharSet ShortInt Character set (see below). MsgType ShortInt Type of MsgData (see below). Area String{255} AreaTag(s) (see below). OrigAddr String{255} Origin address in text format (see below). ReplyAddr String{255} Reply linkage information (see below). FromUser String{255} Who wrote this message? ToUser String{255} To whom? Subject String{255} Subject or one filename (for file attaches and requests). Path String{65535} Path (see below). HeadExt String Header extension fields (see below). MsgData String Message data (see below). HeadSize -------- The total size of the message header (all fields except MsgData). Since this is an Integer, the header can't be bigger than 65535 bytes and Path can never reach it's maximum length (how big it can be depends on how big the rest of the header is). Message flags ------------- Bit Flag Meaning ---------------------------------------------------------------------------- 0 Pvt Private message. 1 File File attach message. 2 FileReq File request message. 3 UpdReq File update request message. 4 Direct Do not route this message. 5 Crash High priority mail. 6 Hold Hold for pickup. 7 IMM Immediate mail (highest priority). 8 RRQ Return receipt request (see below). 9 CRQ Confirm receipt request (see below). 10 IRR Is return receipt (see below). 11 Machine Message to a program (see below). 12 NoForCC CC in NetMail, NoForward in EchoMail (see below). 13 Permanent Message must not be purged by age (see below). 14 Foreign Message is from another organization. 15 reserved Reserved for future extension. RRQ, CRQ and IRR ---------------- The RRQ and CRQ flags without IRR is used to ask the software at the final destination to generate a receipt message to the sender. The receipt to an RRQ flagged message must be generated when the message is received, and this receipt must have the IRR and RRQ flags set. The receipt to a CRQ flagged message must be generated when it has been read by its addressee, and this receipt must have the IRR and CRQ flags set. A receipt must have the ReplyAddr and ReplyID fields set as if it was a reply to the original message. It's up to each network's policy to decide if all systems must support this, no system are allowed to support this or all systems are allowed but not forced to support this. Machine ------- The Machine flag indicates that this message is addressed to a program and not a human. A mail processor should store such messages as defined in the "Stored message" section later in this document so the program doesn't have to support 4711 different message base formats. The name of the program must be in the ToUser field. Carbon Copy ----------- The CC flag is used for sending a NetMail message to several nodes. Which nodes is determined by the MsgDest field. If the message is addressed to a Zone Coordinator, the message is destined for all nodes in that zone. If the message is addressed to a Region Coordinator, the message is destined for all nodes in that region. If the message is addressed to a Net Coordinator, the message is destined for all nodes in that net. If the message is addressed to a HUB, the message is destined for all nodes under that HUB. If the receiver of a CC message have other coordinators (or HUBs) below, it must send the message to them as CC's too. If the ToUser field is empty the message is considered to be addressed to the SysOp and the name is taken from the nodelist (or set to "SysOp"). This must be done by the system immediate before the destination (usually the destination's HUB). Permanent --------- This flag indicates that the message must not be removed when old messages is cleaned out. A new message with this flag automatically erase an old message with the same data in the Subject field. If you want to erase a permanent message without replacing it with a new, then post an empty message with the Permanent flag and the appropriate data in the Subject field. This flag can only be used in EchoMail. OrigAddr -------- This field must contain the true origin address of the message in text format followed by an "@" (40h) and the name of the originating organization. Message ID ---------- The origin address and MsgID fields are used as a unique message identifier. How the MsgID field is generated is left to the implementor, but it must be unique for each message generated on a given node in at least three years. MsgID must normally not be zero, zero is used to indicate lack of Message-ID. With this system these two fields together is unique for every message, and can be used for duplicate detection and other things. Reply linkage information ------------------------- In a reply the ReplyAddr and ReplyID fields must be identical to the origin address, and MsgID fields in the replied message. A message which is not a reply must have these fields empty. Use the ReplyAddr field to check this. Area ---- A list of one or more AreaTags in EchoMail, just a NUL in NetMail. If the list contains more than one tag, they are separated by space. 00h through 1Fh are not allowed in this field (except NUL as the terminator). Path ---- This field contains a series of addresses separated by space. The first system must place its address in this field and every system that routes the message must add its address to this field, Note that the first address in this field and the address in the MsgOrig field doesn't have to be the same. The first address must be the origin address in this format: zone:net/node.point@organization In the other addresses organization, zone, net and node must be omitted if same as the previous address. The other addresses must be in one of these formats: zone:net/node.point@organization zone:net/node.point net/node.point node.point .point .point must be excluded if point is zero, except when .0 stands alone. Example: 1:123/324@FidoNet sends a routed NetMail message to 2:224/546.5; 1:123/300, 1:123/0, 1:12/0, 1:1/2, 2:22/888, 2:22/0, 2:224/0, 2:224/546 and 2:224/546.3 routes the message (in that order). The path field must in this case be "1:123/324@FidoNet 300 0 12/0 1/2 2:22/888 0 224/0 546 .3" when 2:224/546.5 receives the message. Note that the message is routed via a point, that might not be so common but it's technically possible. In EchoMail, the Path field can also contain addresses suffixed by an exclamation mark ("!", 21h). This means that that node hasn't routed the message, but the message has been sent to him. Similar to SEEN-BY's in FTS-4. However, this should only be used when necessary. Never send an EchoMail message to a node listed in the Path field; regardless if the address is suffixed by an exclamation mark or not. If the message originates in a non-FTN organization, the name of the origin organization prefixed by an "@" (40h) must be placed before the address of the gateway. In such cases there may also be some other FTN addresses before the non-FTN organization name if the message have travelled from an FTN, via a non-FTN and to a FTN again. Example: someone@one.two.three in InterNet sends a routed NetMail message to 1:222/345 in FidoNet via the gateway 1:222/111 and 1:222/300. The path field must in this case be "@InterNet 1:222/111@FidoNet 300" when 2:222/345 receives the message. If there's not enough room to update the Path field (add what you should and have at least two bytes left), put a single "$" (24h) in it instead (and a space between it and the last address). This field has the same format as the PTH line specified in FSC-44 (except the "$"). Read FSC-44 for more information. HeadExt ------- This field contains zero or more NUL-terminated strings. The end of this field can be determined by the HeadSize field, but there must be a NUL after the last string anyway. If there are no strings here, there must not be any NUL either. Each string here is a header extension field. A header extension field contains a keyword followed by a space and some data. A header extension field can contain only a keyword with no data, in such case a space must not be used either. Inter-organization messages --------------------------- To send a message to an other organization place the final address in text format followed by an "@" (40h) and the name of the destination organization in a header extension field with the keyword "DEST" and address the message to a gateway. If the message must travel over more than one organization-border, the addresses to each gateway can be pointed out by header extension fields with the keyword "ROUTE". ROUTE fields contains addresses in the same format as DEST lines. The order of the ROUTE fields is important and the DEST field must be after the ROUTE field(s). When the gateway pointed by a ROUTE field processes the message it must change "ROUTE" to "ROUTD". Note: The gateway from your organization are pointed out by the MsgDest field so there must not be a ROUTE field for that gateway. Example: You are 34:65/12@StrangeNet and want to send a message to someone@one.two.three in InterNet. StrangeNet hasn't got any gateway to InterNet, but FidoNet has gateways to both StrangeNet and InterNet. In that case the message may have the following header extension fields: ROUTE 1:1/20@FidoNet DEST someone@one.two.three@InterNet Messages from other organizations (NetMail or EchoMail doesn't matter) must have their origin addresses (including @organization as mentioned before) stored in the OrigAddr field and the Foreign flag set. The MsgOrig field in the header must contain the address of a bidirectional gateway. The MsgID field must be generated using the message-ID in the originating organization, if that is possible. If the originating network doesn't have any message-ID that can be used to create the MsgID field, the MsgID field must be set to zero. If two identical messages (with identical message-ID) enters an FTN organization at two different gateways, they should get the same MsgID. The same counts for ReplyID. ReplyAddr must also be transferred if present in the original message. If the other organization uses a message-ID format that can't be transparently mapped to MsgID, a ORIGID header extension field is created with a direct copy of the original message-ID (both address and serial number). The same is done with reply-linkage information in a ORIGREF header extension field. This information is used if the message should be gated back to the original network. Example with message from InterNet: ORIGID <abcd43kxz@somewhere.org> ORIGREF <bcdef43gkrt@elsewhere.org> Message routing --------------- The transport layer (including intermediate systems (systems handling in-transit messages)) are not allowed to change anything in the MsgDate, MsgID, ReplyID, MsgLength, CharSet, MsgType, OrigAddr, ReplyAddr, FromUser, ToUser, Subject and MsgData fields. No flags may be changed by the transport layer except the Foreign flag at gateways. The only exception is if the message have to be converted to another packet type. An intermediate system must update the Path field, both in NetMail and EchoMail. In EchoMail, the Area field may be changed. An intermediate system are allowed to add or change header extension fields. An intermediate system can add a header extension field with the keyword "Via". The Via field contains information about the system and when it processed the message. It's up to each network's policy to decide if this is allowed in NetMail, EchoMail or both. If the header is changed, the HeadSize field must be updated. EchoMail -------- This section lists the changes to FTS-4 needed to implement EchoMail with this packet format. SEEN-BY lines are not to be used. The Path field is used instead of PATH lines. The Area field is used instead of the AREA line. The MsgOrig and OrigAddr fields must contain the origin addresses of the message. A unaddressed EchoMail message must have the ToUser field empty, unless there is a good reason to put something there. The MsgDest field must be used as in NetMail. If you receive a EchoMail message not addressed to any of you addresses, you must handle it as in-transit NetMail. This makes it possible to route EchoMail through nodes not connected to the area. This can be performed inter-organization as well. Since the lack of complete SEEN-BY's, an EchoMail processor must use the single-pass technique. That means that it must export a message to all its downlinks at the same time as it tosses the message, before it goes into the local message base (it doesn't even have to get there if the area is pass-through). With this technique the tosser can do what ever it wants with the message before it stores it in the local message base since the message will never get out of it again (except when a rescan function is invoked, but the requesting system will have to stand that). A system which receives an EchoMail message with the NoForward flag set must not export that message. This flag is indented for EchoMail processors with a rescan function, all rescanned messages must have the NoForward flag set. The tear and origin lines have no technical meaning (the origin address is stored elsewhere) so no program is forced to create them. Additionally, a program must not require them. To distinguish EchoMail from NetMail, check the Area field. NetMail messages have empty Area fields. Cross-posting a message in several areas is preformed by storing multiple AreaTags in the Area field. If some of your downlinks don't have all the areas listed in the Area field then you must remove the AreaTags for the areas they don't have. All copies must still have the same MsgID. If you receive several messages with the same MsgID but different Area then you must join them into a single message and put all different AreaTags in the Area field (but not the same twice). Compare all fields in the header up to and including Subject (except Area) and not only MsgID. When preforming dupchecking with MsgID, compare all fields in the header up to and including Subject and not only MsgID. CharSet ------- Value Character set ---------------------------------------------------------------------------- 1..99 ISO 8859 (8 bit) 100 ISO 10646 16 bit (Unicode) 101..149 Reserved for other 16 bit character sets 150 ISO 10646 32 bit 151 PC8-437 (The Original) 152 PC8-850 (Multilingual) 153 PC8-852 (Slavic) 154 PC8-860 (Portuguese) 155 PC8-863 (Canadian-French) 156 PC8-865 (Nordic) 157..199 Reserved for other 8 bit character sets 200..255 reserved PC8 means IBM's codepage (8 bit). CharSet=1 means ISO 8859-1 (Latin-1), CharSet=2 means ISO 8859-2 and so on. CharSet concerns FromUser, ToUser and Subject. It may also concern MsgData. Combined characters are built by 02h followed by a modifier and the base character. If you can't display the combined character, just ignore the modifier and display the base character unmodified. A combined character contains three bytes and are counted as three bytes. See FSC-51 for more information on this. There are a few things in FSC-51 that doesn't apply to this format: this method is allowed on all 8 bit character sets, not only Latin-1, and the byte values 80h through 9Fh are allowed. This can only be used for 8 bit character sets. Combined character can be used in FromUser, ToUser and Subject, it may also be allowed in MsgData. MsgLength, MsgType and MsgData ------------------------------ If the MsgData field contains nothing (MsgLength is zero), it's an empty message. An empty message must be forwarded and routed as any other message. An empty message must not be stored in the local message base (but read about the Permanent flag). If MsgType is zero, MsgData is defined by the "Message text" part of this document. If MsgType is non-zero, the MsgData field is not defined by this document. Message text ============ This part defines the presentation layer. The CharSet field in the header concerns the message text except binary extension fields. Combined characters can be used in the message text except in binary extension fields. If a 16 or 32 bit character set is used all characters are two or four bytes so all control codes must be prefixed by 00h or 00h 00h 00h. The bytes alone can be in printable character. The length of binary extension fields is still counted in bytes. All 256 byte values (including NUL) are accepted in the message text. 00h through 1Fh are control codes and must not be used to represent printable characters. A paragraph (also called "physical line") is ended with CR. 8Dh (sometimes called "Soft CR") must not be used as a line separator since it represents a printable character in some character sets. 01h first at a physical line (i.e. the first character of the message text or the character immediately after a CR) means that this line is a extension line. See below for more information. 00h first at a physical line (except in another binary extension field) indicates the start of an long binary extension field. 00h is followed by a LongInt and up to almost 4 GigaBytes of binary data. A binary extension field is not terminated in any way, it's length is stored in the LongInt (the 00h and the LongInt itself is not included in the length). 15h first at a physical line (except in another binary extension field) indicates the start of an short binary extension field. 15h is followed by a ShortInt and up to 255 bytes of binary data. A binary extension field is not terminated in any way, it's length is stored in the ShortInt (the 15h and the Integer itself is not included in the length). Attributes: 03h = Turn off all attributes 04h = Bold 05h = Italic 06h = Underline 17h = All caps (lower case letters as small caps) 19h = Subscript 1Ah = Superscript 1Ch = Blinking 1Dh = Inverted 1Eh = Concealed Subscript automatically turns off superscript and vice versa. All other attributes can be used together. If an attribute already is on, it's turned off by the same code again. If your system doesn't support an attribute, just ignore it. CR turns off all attributes. To quote a message begin a physical line with 1Fh followed by the initials of the person you are quoting and another 1Fh. Use just two 1Fh if you don't want any initials. Successive quoting is marked with more 1Fh:s immediate after the two first ones. A quote is ended with CR. It's up to each implementor to decide how to display quoted text, but it must be possible to distinguish it from normal text. It's recommended to display it in another color if possible. You must translate the quote to the same character set used in the rest of the message. The last paragraph of the message text must be terminated by a CR (i.e. the last character of the message text must be a CR unless a binary extension field is last). Extension lines --------------- An extension line contains a keyword followed by a space and some data. A extension line can contain only a keyword with no data, in such case a space must not be used either. An extension line is ended like a paragraph. Extension lines may not contain 00h through 1Fh. If you want binary data, use a binary extension field instead. Extension lines are control information any should not be displayed to the user. If you want "hidden lines", then use the Concealed attribute rather than an extension line. Information that is relevant for the transport layer must be placed in a header extension field rather than in an extension line. Extension lines can be useful for describing a binary extension field. Currently defined extension lines --------------------------------- A extension line with the keyword FILE is used for sending files embedded in the message text. The keyword is followed by a eight digit uppercase hexadecimal number containing a TimeStamp for the file, a space and the filename. The filename can contain all chars valid for extension lines (including space) and must be at least 1 char and at most 255 chars. The file is stored in a binary extension field immediate after the FILE line. Examples: FILE AB34F657 FILE.EXT FILE 1234ABCD This is a filename Stored message ============== This is a description of the application layer, the local message base. It is recommended to use this format to store NetMail messages locally. It can also be used for EchoMail, but that's not necessary. Each message is stored as a single file which must have a hexadecimal number from 00000001 to FFFFFFFF as the base name and the extension MS3 (to distinguish from TYPE-2 stored messages). Messages can be stored in two ways. Either the same way as TYPE-2 with one directory for each area (including NetMail), in that case the Area field must be empty in all messages. Or all messages in the same directory with the area field used to indicate what/which area the message belongs to (or empty to indicate NetMail). The second format is useful if you normally store EchoMail in an other format and only use this format for messages with the Machine flag. Field Type Description ---------------------------------------------------------------------------- SRdate TimeStamp When the message was sent or received (see below). ReplyTo LongInt Message which this replies (see below). Reply1st LongInt First reply to this message (see below). ReplyNext LongInt Next reply to this message (see below). LocalFlags Word Local flags (see below). Cost Integer Cost of sending in the lowest currency unit. HeadSize Integer Size of the message header (see below). MsgFlags Word Same as packed message. MsgDate TimeStamp Same as packed message. MsgID LongInt Same as packed message. ReplyID LongInt Same as packed message. MsgLength LongInt Same as packed message. MsgOrig Address Same as packed message. MsgDest Address Same as packed message. CharSet ShortInt Same as packed message. MsgType ShortInt Same as packed message. Area String{255} AreaTag(s) (see above). OrigAddr String{255} Same as packed message. ReplyAddr String{255} Same as packed message. FromUser String{255} Same as packed message. ToUser String{255} Same as packed message. Subject String{255} Same as packed message. Path String{65535} Path (see below). HeadExt String Same as packed message. MsgData String Same as packed message. HeadSize -------- The total size of the message header (all fields except MsgData). Since this is an Integer, the header can't be bigger than 65535 bytes and Path can never reach it's maximum length (how big it can be depends on how big the rest of the header is). Please note that this won't contain the same value when the same message is placed in a packet because SRdate, LocalFlags, ReplyTo, Areply and Cost isn't included then and Area might not have the same length. Local flags ----------- Bit Flag Meaning ---------------------------------------------------------------------------- 0 Local Message is created locally. 1 InTransit Message is not destined for this system. 2 Orphan Unknown destination. 3 KillSent Remove message after it has been sent. 4 DelSent Delete attached file(s) after they have been sent. 5 TruncSent Truncate attached file(s) after they have been sent. 6 Sent Message has been sent. 7 Read Message has been read by the SysOp. 8 Rcvd Message has been read by its addressee. 9 Lock Lock the message for further access. 10 DontSend Do not send the actual message but send attached files, make file requests and poll the destination if necessary. 11 unused Reserved for future extension. 12 unused Reserved for future extension. 13 unused Reserved for future extension. 14 unused Reserved for future extension. 15 unused Reserved for future extension. Note that "sent" also can mean "picked up". SRdate ------ The meaning of this field is determined by the Local and Sent flags. Local but not Sent: Meaningless, must be zero. Local and Sent: When the message was sent or picked up. Not Local: When the message arrived to this system. Reply linking ------------- The ReplyTo field points at the message which the current message replies. The Reply1st field points at the first reply to the current message (the reply with lowest number); a messages with no replies must have this field set to zero. The ReplyNext field points at the next reply the same message as the current message; this field must be zero if there are no more replies to the message the current message. A message which is no reply must have the ReplyTo and ReplyNext fields set to zero. With this system there is possible to keep track of an unlimited number of replies to any message. Path ---- If the local flag is set, this field must be empty. It's the mail processors responsibility to insert the address here when placing the message in a packet. When the mail processor tosses an incoming message, it must copy the Path field from the packet. Control file ------------ In each message directory (both EchoMail and NetMail) there must be a file named LASTREAD. When an area is renumbered, this file must be updated. Field Type Description ---------------------------------------------------------------------------- LastRead LongInt The last message read. HighRead LongInt The highest message read. HighWater LongInt High-water mark (only used with EchoMail). This field is used by the EchoMail processor to store the number of the last scanned message. Notes ===== If you are implementing this and something confuses you, ask someone who knows. Please don't guess how it should be. Credits ======= Thanks to Krister Hansson-Renuad, Johan Olofsson, Nils Hammar, Jonas H÷gstr÷m, Mats Gefvert and others who have helped me developing this.