dplatt@teknowledge-vaxc.ARPA (Dave Platt) (09/05/87)
The following text describes some proposed enhancements to the
MacBinary file-transmission standard; authors of terminal emulators
and file-transfer utilities should be aware of the changes so that
their implementations of MacBinary will remain up-to-date.
A new version (2.0) of FreeTerm has been implemented to support the
new flavor of MacBinary. As soon as I get a chance, I'll download it
from my local BBS and post it to comp.binaries.mac for distribution.
----- cut here ----- cut here -----
These are the new changes to the MacBinary Standard, as generally agreed
upon in the MacBinary II Conference 6/21/87, and as changed in the followup
conference 6/28/87. Revised 7/24/87 to reflect suggestions and clarifications
that came later, and to include all necessary information needed from the
original MacBinary standard document to implement MacBinary II.
The new standard will be very similar to the original MacBinary standard as
described in MACBIN.STD, DL3 of AppDev. (Reading the original standard is
recommended for a full understanding of implementation and philosophy
behind the MacBinary I and II formats.) The binary format consists of a
128-byte header containing all the information necessary to reproduce the
document's directory entry on the receiving Macintosh; followed by the
document's Data Fork (if it has one), padded with nulls to a multiple of
128 bytes (if necessary); followed by the document's Resource Fork (again,
padded if necessary). The lengths of these forks (either or both of which
may be zero) are contained in the header.
The format of the header for MacBinary II is as follows:
Offset 000-Byte, old version number, must be kept at zero for compatibility
Offset 001-Byte, Length of filename (must be in the range 1-63)
Offset 002-1 to 63 chars, filename (only "length" bytes are significant).
Offset 065-Long Word, file type (normally expressed as four characters)
Offset 069-Long Word, file creator (normally expressed as four characters)
Offset 073-Byte, original Finder flags
Bit 7 - Locked.
Bit 6 - Invisible.
Bit 5 - Bundle.
Bit 4 - System.
Bit 3 - Bozo.
Bit 2 - Busy.
Bit 1 - Changed.
Bit 0 - Inited.
Offset 074-Byte, zero fill, must be zero for compatibility
Offset 075-Word, file's vertical position within its window.
Offset 077-Word, file's horizontal position within its window.
Offset 079-Word, file's window or folder ID.
Offset 081-Byte, "Protected" flag (in low order bit).
Offset 082-Byte, zero fill, must be zero for compatibility
Offset 083-Long Word, Data Fork length (bytes, zero if no Data Fork).
Offset 087-Long Word, Resource Fork length (bytes, zero if no R.F.).
Offset 091-Long Word, File's creation date
Offset 095-Long Word, File's "last modified" date.
Offset 099-Word, length of Get Info comment to be sent after the resource
fork (if implemented, see below).
*Offset 101-Byte, Finder Flags, bits 0-7. (Bits 8-15 are already in byte 73)
*Offset 116-Long Word, Length of total files when packed files are unpacked.
This is only used by programs that pack and unpack on the fly,
mimicing a standalone utility such as PackIt. A program that is
uploading a single file must zero this location when sending a
file. Programs that do not unpack/uncompress files when
downloading may ignore this value.
*Offset 120-Word, Length of a secondary header. If this is non-zero,
Skip this many bytes (rounded up to the next multiple of 128)
This is for future expansion only, when sending files with
MacBinary, this word should be zero.
*Offset 122-Byte, Version number of Macbinary II that the uploading program
is written for (the version begins at 129)
*Offset 123-Byte, Minimum MacBinary II version needed to read this file
(start this value at 129 129)
*Offset 124-Word, CRC of previous 124 bytes
*This is newly defined for MacBinary II.
All values are stored in normal 68000 order, with Most Significant Byte
appearing first then the file. Any bytes in the header not defined above
should be set to zero.
The original MacBinary format was amended to include the sending of the FCMT
(Get Info comment) after the resource fork was sent, if the length for such
comment, given in offset 99, is not zero. To the best of our knowledge, no
program has implemented this feature, due to Apple's stated position that no
program should read or write these comments. The definition remains in
MacBinary II, so that should Apple ever provide a documented way of reading and
writing these comments, terminal programs will be able to take advantage of
this feature.
All Finder flags and information would be uploaded, however, a downloading
program should clear the Finder flag bits of
0 - Set if file/folder is on the desktop (Finder 5.0 and later)
1 - bFOwnAppl (used internally)
8 - Inited (seen by Finder)
9 - Changed (used internally by Finder)
10 - Busy (copied from File System busy bit)
Also, fdLocation and fdFldr should be zeroed
To determine if a header is a valid MacBinary header, check bytes 0 and 74 to
be both zero. If they are both zero, either (a) the CRC should match, which
means it is a MB II file, or (b) byte 82 is zero, which means it may be a MB I
file. (Note that, at the current version level, byte 82 is kept zero to
maintain compatibility with MacBinary I. If at some point the MacBinary
versions change sufficiently that it is necessary to keep MacBinary I programs
from downloading these files, we can change byte 82 to non-zero.)
If the header is a MB II header, the program will check the minimum version
byte, to see if it knows enough to decode the file. If the minimum version in
the header is greater than the version that the terminal program was written
for, it will download the file as pure XModem (creating a "TEXT" file) and
notify the user that conversion is needed because the MacBinary version was too
high.
If the header does NOT represent a valid MB II header, the program must at
minimum check byte 82 to be zero--if it is not zero, the file is not a MB I
file. It is possible to write a much more robust routine, by checking the
following:
Offsets 101-125, Byte, should all be 0.
Offset 2, Byte, (the length of the file name) should be in the range of 1-63.
Offsets 83 and 87, Long Word, (the length of the forks) should be in the
range of 0-$007F FFFF.
If any of these tests fail, the file is not a valid MacBinary file. It may
still be desirable to distinguish between text files and foreign binary files
(for stripping line feeds or similar helpful acts). Some tests that would
prove useful include:
A quantity of bytes in the first block with the high bit set would point to
a binary file (though this could be fooled by files with many extended
ascii characters, such as generated by the option key on a Mac).
A large quantity of zero bytes (nulls) would also point to a binary file.
------------------------------------------------------------------------
This proposal was adopted at two conferences attended by representatives from
CompuServe, Delphi and BIX networks, and many terminal software publishers. A
partial list of those participating is:
Peter Olson/ICONtact
Larry Loeb/BIX
Neil Shapiro/Maug
Mark Hagerman
Michael Pester
William Bond
Bill Steinberg
Don Brown
Bill Davis
Jean Hess
Scott Watson
Clay Maeckel
Harry Chesley
Jack Brindle
Raines/BMUG
Harry Conover
Chris Allen/Dreams of the Phoenixdaveb@geac.UUCP (Brown) (09/08/87)
In article <16387@teknowledge-vaxc.ARPA> dplatt@teknowledge-vaxc.ARPA (Dave Platt) writes: >The following text describes some proposed enhancements to the >MacBinary file-transmission standard; ... >The format of the header for MacBinary II is as follows: > Offset 000-Byte, old version number, must be kept at zero for compatibility >[discussion of how to infer version with 0 version number] You perhaps misunderstand the meaning/use of a version number. Back in the days of the ARPANET and the Project MAC, they were invented to provide a transparent means of allowing programs to evolve. If there is only one version number (the common case), it is incremented to indicate to an old program that this is a new-format file (or packet!) and that it needs to execute its "exceptional case" code. Often this meant is was to complain to a human and, in the case of translators, refrain from translating things into broken bits. If the version number is not incremented, the program can process the file/packet, ***ignoring** present-but-unimplemented fields and still produce a valid packet/file. If there is a two-part version number (rare) or there is a negotiation procedure (common on ARPANET/in Kermit) the program can use the additional information to read and interpret present-but-unimplemented fields, since it has just been "told" which ones are in use. Please do not publish a standard which does not have a mechanism for evolution and change. The world will insist on changing... In the case of MacBinary, you may want to release a standard and a suite of programs which: 1) ignore present-but-unimplemented fields 2) provides a few for extension 3) has and uses version number of 0, but... 4) defines version 1 and can read all version 1 files 5) does not send version 1 until 1 january 19XX, then switches automagically. Please feel free to contact me by mail, I will be glad to lend a hand (outside of working hours (:-})) at helping you improve the standard without breaking existing software. --dave (notice the quote below) c-b -- David Collier-Brown. {mnetor|yetti|utgpu}!geac!daveb Geac Computers International Inc., | Computer Science loses its 350 Steelcase Road,Markham, Ontario, | memory (if not its mind) CANADA, L3R 1B3 (416) 475-0525 x3279 | every 6 months.