As the packet file specification was originated on x86 systems, which are little endian, problems come up with reading/writing such files on big-endian systems. Use the explicit < (little endian) modifer for the pack/unpack functions? Or change which format codes are being used? Reference also the Byte-order-modifiers section in perlpacktut.
Use the same key names for the message hash for in the read and write packet functions, so as to avoid confusion; match them up as necessary to the names of the fields in the ftsc document.
The $Origin, $seen_by, & $Path variables at Lines 322-324 in the write_ftn_packet function of Packet.pm. Should that creation of an origin line really be in the packet file creation? Or is that something that should be getting created in the message body and getting passed that way? It also hard codes "1" as the origin point number, when that should be OrgPoint. Same for the $seen_by and $Path variables; they are part of the messages, but should not they already be in the message bodies being passed to the write_ftn_packet function?
Fix the possible ftsc date related issues in the course of creating and using an ftsc_date function.
Add a separate function for reading a packet header when passed the name and path of an FTN packet file. Also add other functions as necessary: like one for generating a serial number, one for generating a name for a standard packet file, one for generating an ftsc standard date.
In order to provide Object Oriented functions for writing an FTN Packet, start with a merge of the code FTN::Pkt into FTN::Packet.
The attribute word shouldn not be thrown away? Just put it as is in the hash bing returned.
Areas & Tearline hash references should be in the packet_info hash or a message hash? Certainly there is an area field included in the msg hash being returned by the read_ftn_packet.
When doing, for instance, two messages: two messages were making it into the packets but they are both the text of the second message. The "foreach" line at 354 does appear to work (instead of the two commented out versions of initiating a loop at 355-356) because two messages are being put into the packet file, but the message body does not appear to be makeing it in. (Do the two commented out lines after that need to be kept?) The line at 354 should be bringing the Body of the message in; no apparent error when running a test script but is that line correct?
Also the serial number for the MSGID is coming up as the same for both messages. When code a better way of getting a 8 char serial number to use, add it as a function to the module?
The writeup and variable usages for the packet format per FTS-0001 could be rewritten to more closely match the current documentation, esp. the listed order.
Change the name of the appropriate fields in the packet related hashes to being named after how they are named in the FTS-0001 documentation.
A second tear line of just "---" is showing up in the body of a message.
A period is showing up at the end of the ftscdate variable; is that actually how the ftscdate is in the packet or is that an artifact of the code?
Regardless of if they are to be used or not, the $packet_version, $attribute, and $cost information being pulled from the message headers should be returned in the message hash like the rest of the information.
Change the name of the appropriate fields in the packet related hashes to being named after how they are in the fts-0001 documentation, possibly including how they are capitalized.
The zone number information is not being pulled from what is available in the packet; need to set the from_node & to_node variables instead of assuming zone 1. It is in the packet header but is not being read in the current code, so at least that part of the packet header needs to be read.
Make the message Reply IDs availabe in the same way as the Msg IDs? (Are they in use?)
Provide the message control info as a hash?
Actually process the packet header instead of just skipping over it, as there are least some informtion (like zone numbers) are needed.
A separate function read and return only the packet header information as hash.
Can or should something else be used for testing the correct number of messages in a packet besides is_deeply()?
Robert James Clay,
<jame at rocasa.us>
Copyright 2014 Robert James Clay, all rights reserved.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.