This document describes the format of L and D/E format directories.

For both formats the directory is arranged as follows: DirHeader
Entries[n]
DirTail

The header and tail contain information about this directory, and the entries are the directory entries.

The two directory formats have the same DirHeader:

Name            Bytes
StartMasSeq     1       Update sequence number to check dir start with dir end
StartName       4       'Hugo'; 'Nick'

The two directory formats have mostly the same entry format:

Name                    Bytes
DirObName               10
DirLoad                 4
DirExec                 4
DirLen                  4
DirIndDiscAdd           3

The DirTail formats are quite different: Old:

Name            Bytes
OldDirLastMark  1       0 to indicate end of entries
OldDirName      10      Directory name
OldDirParent    3       Indirect disc address of parent
OldDirTitle     19      Directory title
Reserved        14      unused (padding)
EndMasSeq       1       To match with StartMasSeq
EndName         4       'Hugo'; 'Nick'
DirCheckByte    1       Check byte on directory

NewDirLastMark  1       0 to indicate end of entries
Reserved        2       unused (padding)
NewDirParent    3       Indirect disc address of parent
NewDirTitle     19      Directory title
NewDirName      10      Directory name
EndMasSeq       1       To match with StartMasSeq
EndName         4       'Hugo'; 'Nick'
DirCheckByte    1       Check byte on directory

The directory sizes are:
Old:
&500 (1280) bytes/47 entries
New:
&800 (2048) bytes/77 entries

There should be an exact number of entry slots between the headers and tails of both formats of directory (hence the padding). The last entry is indicated by there being a 0 in the first byte of the next entry's DirObName (hence xxxDirLastMark is there to indicate end of directory when the directory is full). DirNames are control character terminated and may be the full length of the field they sit in (in which case there is no terminator).

Calculating StartMasSeq and EndMasSeq:
StartMasSeq and EndMasSeq are there to check whether the directory was completely written out when it was last written out. For an unbroken directory they are always equal, and are increased by one (wrapping at 255 back to 0) whenever the directory is updated. This means that if the writing of the directory was stopped halfway through then the start and end master sequence numbers will not be the same, and so the directory will then be identified as broken. Their values should equal each other, but, apart from that, they can be anything.

Calculating DirCheckByte:
This is an accumulation of the used bytes in a directory. The used bytes are all the bytes excluding the hole between the last directory entry and the beginning of the structure at the tail of the directory. The generation of the check byte is best described as an algorithm:

• Starting at 0 an accumulation process is performed on a number of values. Whatever the sort of the value (byte or word) it is accumulated in the same way. Assuming r0 is the accumulation register and r1 the value to accumulate this is the accumulation performed:
  EOR r0, r1, r0, ROR #13
  
• All the whole words at the start of the directory are accumulated. This will leave a number of bytes (0 to 3) in the last directory entry (or at the end of the start structure in a directory if it's empty).
• The last few bytes at the start of the directory are accumulated individually.
  
• The first few bytes at the beginning of the end structure of the directory are accumulated. This is done to leave only a whole number of words left in the directory to be accumulated.
• The last whole words in the directory are accumulated, except the very last word which is excluded as it contains the check byte.
• The accumulated word has its four bytes exclusive-ored (EOR) together. This value is the check byte.