nulib2/library/nulib324doc.txt

NuFile eXchange (NuFX) Archive Utility
NuLib v3.2 Documentation

By Andy McFadden



Updated September 23, 1992













Overview
--------

NuLib is a shell-based NuFX archive utility, based loosely on "ARC"
for the IBM PC and "ar" under UNIX.  It allows you to perform certain
operations on the same archives used by ShrinkIt, including view
archive contents, add to archive, extract from archive, and delete
from archive.  In addition, it will list and unpack files from Binary
II archives.

This program is primarily targeted at users of non-Apple II computer
systems, from IBM PC compatibles to Sun workstations to Amdahl
mainframes.  It will also be of use to people who use a GS/OS shell on
the Apple //gs, such as APW, ORCA, or ECP-16 (although it is
considerably slower than ShrinkIt).  NuLib may require more than 256K
of free RAM to function properly.  Apple //gs users can use YankIt
instead, which is much smaller and many times faster (though
considerably less powerful).


Background
----------

For our purposes, an archive is simply a collection of one or more
files stored as a single file.  This allows groups of related files to
be stored under a single filename, reducing directory clutter, and
makes it possible to transfer groups of files without the use of a
batch transfer utility.

To reduce the space required to store the archives and the time it
takes to transfer archives, most popular archiving programs
automatically compress files as they are stored.  Two popular
compression methods are Huffman (variable-size codes are used, with
smaller codes assigned to bytes that appear frequently) and
Lempel-Ziv-Welch (LZW; finds common substrings in the file).

Popular archiving programs include "BLU" (Apple II), "ShrinkIt" (Apple
II), "GS/ShrinkIt" (Apple IIgs), "ARC"/"PKARC" (MS-DOS), "PKZIP"
(MS-DOS), "lharc" (MS-DOS/Amiga), "zoo" (MS-DOS/Amiga), "PackIt"
(Macintosh), "StuffIt" (Macintosh), "tar" (UNIX), and "ar" (UNIX).
UNIX archivers don't generally compress files; a separate program
(called "compress" of all things) is usually used.  NuLib works with
archives that ShrinkIt produces (NuFX).


New Features/Bug Fixes
----------------------

Major changes since last release:

+ Compression speed improvement
+ Can add a file as if it were a disk image


How to Use NuLib
----------------

Usage: nulib option[suboption] archive [filespec1] [filespec2] [...]

"option" is a single character, as specified below.
"suboption" is one or more characters, which modify the performance of
the option (see each specific option for details).
"archive" is the name of a NuFX archive.  It doesn't need to exist for
the add, create, and move options.
"filespecN" is the name of a file, either on disk or in the archive.
More than one file may be named on the command line.

The option/suboption string may be entered in upper or lower case, and
may be preceeded by a hyphen (ex: "nulib -XT0 ..." is equivalent to
"nulib -xt0 ..." and "nulib xt0 ...").  Suboptions may be entered in
any order.

Side note: Under APW, both the archive filename and all file
specifications relating to files ON DISK are expanded by NuLib (device
names, wildcards, etc).  Under UNIX, all filename expansion and
wildcards are handled by the shell or the kernal.  The filenames of
files in the ARCHIVE do not undergo wildcard expansion on any system.

Examples:
$ nulib av foo.shk file1 subdir/file2
Adds "file1" and "subdir/file2" to "foo.shk", listing each file as it
is archived.
$ nulib av foo.shk subdir
Recursively adds all files in subdir and all files in subdirectories
of subdir.
$ nulib mv foo.shk =.doc    (APW)
$ nulib mv foo.shk *.doc    (UNIX)
$ nulib mv foo.shk *.doc    (MS-DOS)
Adds all files with the suffix ".doc" to the archive "foo.shk",
listing them as it goes, then deletes them from the disk.
$ nulib d+ foo.shk work/
Deletes all files in "foo.shk" that start with "work/"; this would
likely be used to delete all files archived from the directory "work".
No output is produced.


The various options are:

*** Add/Append files:
    nulib A[V][U|C<val>|S<val>][R][F<type>[/<aux>]][D] archive files
Quickly appends the specified files to the end of the archive.  This
option does not scan the archive to see if the files already exist,
and will create a new archive file if one is not found.  This option
corresponds roughly to N)ew archive or A)dd files on the ShrinkIt
menu.

If you add files from the current directory or a subdirectory of the
current directory, then the entire partial pathname is stored (ex:
"nulib av fubar.shk foo/bar/filename" will store "foo/bar/filename" in
the archive "fubar.shk.")  If you add files from directory that is not
a subdirectory of the current directory, only the filename will be
stored (ex: "nulib av fubar.shk ../zip/bang/filename" will store
"filename" in the archive).  Unfortunately, this does not work under
UNIX, since filenames are expanded by the kernel instead of by NuLib.

Wildcards are allowed, and subdirectories are recursively descended
(the depth is limited only by the maximum length of a pathname and the
number of files in the subdirectories), unless the R suboption is
used.  The R suboption will prevent subdirectories from being
processed, so that "nulib avr fubar.shk *" under UNIX will add all
files in the current directory, but will not descend into any
subdirectories.  The V suboption displays the filenames as they are
added.

The U suboption adds a file without compressing it.  The C suboption
allows you to insert a file in the archive with a specific compression
algorithm (specified by the argument "value", which must immediately
follow the suboption - no spaces between.  Ex: "nulib ac5v fubar.shk
file1" would store "file1" in "fubar.shk" using 16-bit UNIX-style LZW
compression).  "value" corresponds to the appopriate thread_format as
specified in the section on compression methods.

The S suboption does not compress the file, but will store it as if it
were by setting the thread_format field to "value".  One use for this
is adding compressed files without having to uncompress them first
(ex: "nulib as1 foo.shk file1.qq" adds a squeezed file).  This
suboption could conceivably cause a NuFX extractor to crash; please
don't use it unless you are sure of what you are doing.  Note also
that some compression methods (like ShrinkIt LZW) require both the
"compressed length" and "uncompressed length" values to be correct;
use of this suboption will (incorrectly) set them both to the current
file length.

Only one of U, C, or S may be specified.

The F suboption allows specification of the filetype attribute.  It
should be three characters long, and will be matched against a list of
standard ProDOS file types.  Common ones are "BIN" (binary), "TXT"
(text), "SRC" (source code), "OBJ" (object code), "EXE" (executable),
and "NON" (typeless, the default).  Both upper and lower case letters
will match.

If the filetype is followed by a "/", then the auxtype field will also
be set.  NuLib expects the auxtype to be a four-byte hexadecimal
number; entering less than four digits could cause the auxtype to be
misinterpreted.  A good example is storing APW C source files: "nulib
cvfSRC/000a nulib.shk file1.c file2.c ...".

If the D suboption is given, the entry will be tweaked to make it
appear that the file is actually a disk image.  This may be useful
somewhere.  Don't use it if you don't know what you're doing.

Adding comments is not currently supported.


*** Binary II Operations
    nulib B[X][V][T<value>][I] binary2-archive [archived-files]
This option is included for compatibility with the older Binary II
standard, since it is still in use by many communications programs.

With no suboptions, this will list the information on the archived
files specified by "archived-files".  The X suboption extracts the
named files from the archive, unsqueezing them if necessary.  If the
archived-files parameter is omitted, then all files will be listed or
extracted.  Note that matches are case-independent.

When the V suboption is used in conjunction with the X suboption, the
names of the archived files are printed as they are extracted.  The V
suboption has no effect otherwise.

The T suboption will perform a text substitution as it works; see the
section of conversion of line terminators for more information.

If the filename of an extracted file would conflict with an existing
file, the file will be overwritten and a message will be printed.  If
the "I" suboption is used, then NuLib will give an "overwrite (y/n)?"
prompt.  An affirmative answer overwrites the file; a negative answer
will skip the file and continue processing.

Attempting to perform NuFX operations on a Binary II archive will
fail.  If the file appears to be Binary II format, then a message
indicating this will be printed.  Providing transparent support for
Binary II archives is not impossible, but isn't needed often enough to
be worth doing.


*** Create Archive
    nulib C[V][U|C<val>|S<val>][R][F<type>[/<aux>]][D] archive files
This is identical to the A option, but the "creating archive" message
is suppressed.  This behavior is similar to "ar".


*** Delete from Archive
    nulib D[V][+] archive archived-files
Deletes the named files from the archive.  NuLib scans the archive,
marking all records that match the names in the file specification
(case-independent).  If all files are marked for deletion, then the
archive file itself is deleted.  Otherwise, a new archive is created,
the unmarked records are transferred, and the old archive is deleted.
An error during this process will leave the original archive
unmodified.

Note that this does not require an exact match if the "+" suboption is
used; "nulib d+ fubar.shk foo" will delete "foo", "Foozle",
"food/recipies" and "food/stock."  The V suboption prints the list of
marked records as it works.


*** Freshen Archive
    nulib F[V][U|C<val>|S<val>][R][F<type>[/<aux>]][D] archive files
Updates files in the archive, but doesn't add any new files.  Creates
a new archive file, and either transfers the old record or adds the
file depending on which is more recent.  Only exact filename matches
are allowed (case-independent), including partial paths.  The archive
being updated must already exist.

Wildcards are allowed, and subdirectories are automatically expanded
unless the R suboption is used.  The V suboption displays the
filenames as they are added.  The U/C/S/F suboptions, explained under
A)dd, only apply to files being added to the archive or being updated
(files that aren't updated are left unaltered).

Files that are updated will retain their previous filetype.  New files
will get either the default filetype, the filetype specified by the F
suboption, or the actual filetype (under APW only) in that order.


*** Command Help
    nulib H[NWS]
Displays a help screen.  Three screens are available.

"nulib h" alone displays help on the options.  The N suboption gives
help with numbers; it lists the known compression methods and text
translation types.  The W suboption gives a brief listing of
contributors, and how to contact me.  The S suboption gives help on
the suboptions.


*** Integrity Check
    nulib I[V] archive
Verifies that the archive is intact.  Does not modify the archive in
any way.  The V suboption prints a list of CRCs for each entire record
(this is different from those listed by the TZ option, which are only
for the record headers and (sometimes) threads; this includes not only
the headers but *all* data as well).

Please note that this doesn't do much more than read the file, unless
the record_version is $0002 or greater (which means that the data has
a checksum stored; currently these records are only generated by
GS/ShrinkIt), AND no compression was used.  This merely scans the
records and verifies the header CRCs, NOT the data CRCs.  The main
purpose of the V suboption is to make a list of CRCs that can be sent
along with the archive.


*** Move Files to Archive
    nulib M[V][U|C<val>|S<val>][R][F<type>[/<aux>]][D] archive files
This is identical to the A option, but the files are deleted after
they are added to the archive.  Note that the actual directory files
are NOT deleted, unless they were given distinct record entries.

Care should be taken to avoid trying to M)ove an archive into itself.
The act of adding may (depending on the OS and the archive) go into an
infinite loop creating a huge file, and the coup de grace is when
NuLib then deletes the archive you were adding to.


*** Print an Archived File
    nulib P[V][T<value>][+] archive files
Print the contents of an archived file without extracting it.  Useful
for viewing archived text files without having to actually unpack
them.  Note this only allows viewing of data_threads; resource forks
and disk images will not be displayed, and comments are not shown.  I
take no responsibility for pagination or filtering of funky control
characters...

The V suboption will print the file's name right before it is
extracted.  The + suboption allows you to specify the first part of a
pathname; see the D or X options for details.

The T suboption will perform a text substitution as it works; see the
section on conversion of line terminators for more information.


*** Table of Contents
    nulib T[VAZ] archive
With no suboptions, this lists only the filenames of the archived
files.  Not only does this make it easier to view the archive contents
(the ShrinkIt format filename field is about 20 characters wide; this
is as wide as it has to be), but the output is suitable for
transmission via a pipe to other utilities.

Using the V suboption will make it use ShrinkIt v2.0 output format
(same as using the V option); it is included as a suboption mainly for
people used to "ar".  Using the A suboption will produce a list
similar to the output of ARC or ZOO.

Using the Z suboption will dump everything known about the archive,
including all information in headers, CRCs, relative file position,
sizes of individual threads, etc.


*** Update Archive
    nulib U[V][U|C<val>|S<val>][R][F<type>[/<aux>]][D] archive files
Updates files in the archive, keeping the archived file or the file
listed on the command line, whichever is most recent (or exists).
Unlike freshen, this will add new files as necessary.  Creates a new
archive file, and either transfers the old record or adds the file.
Only exact filename matches are allowed (case-independent), including
partial pathnames.  The archive being updated must already exist.

Wildcards are allowed, and subdirectories are automatically expanded
unless the R suboption is used.  The V suboption displays the
filenames as they are added.  The U/C/S/F suboptions, explained under
A)dd, only apply to files being added to the archive (files that
aren't updated are left unaltered).  Note that the order of files in
the archive will be preserved.

Files that are updated will retain their previous filetype.  New files
will get either the default filetype, the filetype specified by the F
suboption, or the actual filetype (under APW only) in that order.


*** Verbose Archive Listing
    nulib V archive
Lists the archive contents in a format indentical to that used by the
ShrinkIt v2.0 L)ist archive contents option.  Same thing as "-tv".


*** Extract from Archive
    nulib X[V][T<value>][U][I][M][+] archive [archived-files]
    nulib E...
The X and E options are synonymous.  Extract the archived-files from
the archive.  If the file already exists, you are asked if you want to
overwrite it.  If part of a partial pathname does not exist, the
subdirectory will be created.  Omitting the "archived-files"
specification will cause the entire archive to be unpacked.

When files are archived, a pathname separator is stored (e.g., "/" for
ProDOS and UNIX, "\" for MS-DOS, ":" for Mac HFS).  During extraction,
the pathnames are broken down into component file names, converted to
names legal under the current operating system, and then recombined
using the pathname separator for the current OS.  This facilitates
extraction of files archived under any OS, but can lead to filename
conflicts that didn't exist when the files were added (e.g., a UNIX
file that contained a backslash is unpacked under MS-DOS).

If the filename of an extracted file would conflict with an existing
file, the file will be overwritten and a message will be printed.  If
the "I" suboption is used, then NuLib will give an "overwrite (y/n)?"
prompt.  An affirmative answer overwrites the file; a negative answer
will skip the file and continue processing.

Note that extraction does not require an exact match if the "+"
suboption is specified; "nulib x+ fubar.shk foo" will extract "FOO",
"Foozle", "food/recipies" and "food/stock."  This makes it possible to
extract entire subdirectories at a time.  The V suboption prints the
list of marked records.

If comments are present, then the V suboption also prints those as it
extracts.  Using the M suboption will cause the comments to be
printed, but the files will not be extracted.  **NOTE**: the exact use
of the M suboption has not been entirely settled.  Since it may be
desirable to extract comments into a file, future versions of NuLib
may use eXtract to put them in a file and Print to view them.
Comments welcome, caveat emptor, have a nice day.

The T suboption will perform a text substitution as it extracts; see
the section on converson of line terminators for more information.

The U suboption will extract the files without uncompressing them.
This is rarely useful, but is easy to implement and is present for the
curious.


Non-NuFX files
--------------

NuLib will only work with Binary II and NuFX archives.  If you try to
view some other kind of file, you will get an error message and an
indication of what kind of file NuLib thinks it is.  NuLib can
recognize files processed with compress, Zip, Zoo, StuffIt, ar, shar,
GIF, and many others.


Compression Methods
-------------------

The following methods are defined in the NuFX documentation:
(# = method number, Name = method name [abbreviation], Pack? and
Unpack? refer to the ability of NuLib to perform that operation using
the given compression method.

#    Name                             Pack?    Unpack?
00   Uncompressed [unc]                 Y         Y
01   SQueezed [squ]                     N         Y
02   Dynamic LZW I (ShrinkIt) [shk]     Y         Y
03   Dynamic LZW II (ShrinkIt) [sh2]    N         Y
04   12-bit UNIX Compress [u12]         Y         Y
05   16-bit UNIX Compress [u16]         Y         Y

Attempting to use a compression method that does not exist will result
in an error message like "[can't squeeze; storing]...".  This means
that the compression method you requested is unavailable, and it
simply stored the file without compression.

If you try to extract a file that has been compressed with an
algorithm that NuLib is not familiar with, an error message will be
printed and the file will not be extracted.


Converting Line Terminators
---------------------------

Different operating systems use different line terminators in text
files.  The table below shows them for some popular systems:

Operating System         Line Terminator
Apple ProDOS             CR ($0d)
UNIX                     LF ($0a)
MS-DOS                   CRLF ($0d0a)

While NuLib will know what kind of terminators the operating system it
is running under uses, it cannot reliably determine what kind an
archived file uses.  Thus, the terminator used on the system where the
file was created must be specified.

Note that text translation should *not* be performed on non-ASCII
files (non-ASCII means anything other than pure text; word processing
files and executables are two examples of files that should *never*
have text translation performed on them).  Doing so will probably ruin
the file, causing strange errors.  Because of the wide range of files
that NuLib must handle, it is impossible to automatically determine
which files are binaries and which are text.

In order to tell NuLib what format to expect, you have to specify a
value parameter (although it will default to zero if you don't).  The
following examples illustrate their usage:

nulib xt0 ...       Convert from CR   (Apple II -> current system)
nulib xt1 ...       Convert from LF   (UNIX -> current system)
nulib xt2 ...       Convert from CRLF (MS-DOS -> current system)


Shell Variables
---------------

NuLib looks for a shell variable called "NULIBOPT" to get default
values for certain things.  This must be an environment variable
("setenv" using csh, "set" and "export" using sh or APW), and may
contain the following:

     verbose             Default to Verbose output (otherwise Silent).
     type=xxx            Default type for storing files (under UNIX or
                         MS-DOS).  "xxx" should match the 3-letter
                         ProDOS abbreviation (NON, BIN, TXT, SRC).
                         Normally "NON".
     aux=xxxx            Default auxtype for storing files (under UNIX
                         or MS-DOS).  "xxxx" is a four-character hex
                         number (000a, 8002, abcd).  Normally "0000".
     interactive         Default to Interactive mode (prompt before
                         overwriting files).
     compress=n          Set default compression to type n.  This is
                         useful if you prefer 16-bit UNIX compress
                         (which would be compress=5).

Note that the type= and aux= settings do NOT apply when running under
APW.  Files will be stored with their actual filetypes, regardless of
the variable setting.

Also, the 'F' suboption will override these settings.

An example from csh:
  setenv NULIBOPT type=BIN,compress=5,interactive

** WARNING: As of this writing, GS/ShrinkIt 1.0.6 and ShrinkIt v3.4
are not able to extract archived files compressed with UNIX compress.
It is likely that GS/ShrinkIt will be able to in the future, but the
//e version of ShrinkIt will probably never handle them.


Error Handling
--------------

Many errors simply cause the program to exit, leaving the archive in
an uncertain state (which sounds fairly evil).  If you were
extracting, deleting, viewing, or updating when the error occurred,
the worst that can happen is you will be left with a bogus temporary
file in the current directory (something like "nulib.tmp5610").

If you were adding to an existing archive, the files that were there
will be unharmed, but additional files will not appear, and the
archive will be oversized.  This is because the master header block
(which keeps a count of the number of records in the archive) is
written last.

If you were creating a new archive, the file will be guano.  This is
because, as mentioned before, the Master Header Block is not written
until the very end.  Since NuLib identifies NuFX archives by looking
at certain bytes in the MHB, the file will not be identifiable as
NuFX.  Note that the M)ove option is safer than it looks, because
files on disk are not deleted until the archive is safely closed.


Revision History
----------------

NuLib v3.22 (September 1992)
- improved speed of LZW-I compression
- added ability to add files as if they were disks (patches provided
by somebody whose name I've lost)
- updated the list of 3-letter filetype abbreviations

NuLib v3.21 (August 1992)
- minor stuff

NuLib v3.2 (April 1992)
- fixed two bugs in the LZW-II uncompression.
- made it compatible with SysV Expanded Fundamental Types (stuff like
eight byte file offsets on lseek()s).

NuLib v3.11 - v3.14 (November 1991)
- fixed a problem with LZW-II uncompression.
- added fixes to make it compile on a NeXT and with MS C 6.0.
- fixed XENIX directory compatibility problems.

NuLib v3.1 (October 1991)
- added ability to uncompress LZW-II compression (GS/ShrinkIt).
- improved speed of compression routines.
- improved System V compatibility.
- cleaned up code a bit and fixed minor bugs.

Nulib v3.01 - v3.03 (February 1991)
- fixed XENIX problems with includes and libs.
- fixed bug in directory expansion.
- silenced screaming about bad dates.
- fixed glitches in nulib.lnk and nulib.mak
- fixed non-compression bug in ShrinkIt LZW.

NuLib v3.0 (Sep 1990):
- added ability to compress files using ShrinkIt LZW-I compression
(replaces "fake" compression used previously).
- added "compress" parameter to NULIBOPT environment variable.
- added ARCZOO output format for people used to MS-DOS archivers.
- added M suboption for comment ("message") printing.

NuLib v2.3 (May 1990 - not released):
- addition of UNIX compress.
- threw in some benchmarks.

NuLib v2.2 - v2.22 (Apr 1990):
- second release as NuLib (after a brief vacation).
- fixed incompatibility problems with GS/ShrinkIt and ShrinkIt v3.0.
- added some support for comments.
- unpacks disk archives to files (good for the UNIX Apple ][+
simulator).
- fake compression correctly handles files that don't compress (it
used to store them as compressed whether or not they got smaller).
- various minor changes/improvements (more help screens, etc).

NuLib v2.1.1 (Nov 1989):
- first wide distribution as NuLib.
- fixed command processing bugs.

NuLib v2.1 (Nov 1989):
- yet another name change (thanks loads, L&L).
- ShrinkIt LZW uncompression added.
- MS-DOS code added.
- shell variable used for defaults.
- CRLF translation.
- added R, I, and F suboptions.

CShrink v2.08 (Oct 1989):
- altered help screens, some commands.
- added recognition of other kinds of files (compressed, shar, etc).
- switched to table lookups for calculating CRC (much faster).

CShrink v2.07 (Sep 1989):
- Another name change (legal reasons).
- UNIX port completed.
- Binary II operations are fully functional.
- some compression code added (unSQueeze, fake ShrinkIt pack).
- '+' suboption added.
- text translation improved.
- printing of archived files (P option) now works.

NuARC v2.03 (Aug 1989):
- first NuARC distribution (APW executable only).
- added subdirectory expansion.
- added suboption processing.
- replaced buffered I/O (fopen(), fread(), etc) on files with
lower-level read()/write() routines.
- added automatic byte-order determination.
- implemented move, extract all, and update/freshen.
- added Print archived file option.
- added Verbose, Text translation, and Uncompressed storage
suboptions.
- wrote this documentation.

NuARC v2.0 (Aug 1989):
- added archive manipulation routines (EXtract, Add and Delete for
uncompressed archived files).
- added filename-only output format.
- added CRC verification.
- added byte order and data element size checks.
- removed the LAMESEEK option.

NuView v1.2 (July 1989):
- major overhaul of all source code to make it work under APW C.
- new //gs-specific routines added.
- minor alterations to output format.

NuView v1.1 (June 1989):
- major rewrite of the way archives are read (had problems with
machines requiring word-alignment of data).
- added thread file position storage to internal archive structure.
- fixed non-(void) sprintf() bug.

NuView v1.0 (May 1989):
- initial release.
- works only on a Sun 3/50 running BSD UNIX.


Limitations
-----------

NuLib works just fine with records containing more than one thread
(i.e., comments and resource forks).  However, while comments can be
printed, they can't be added.

The big problem this program has is speed.  Since it is meant to be
portable first and efficient second, it won't run as fast as something
like ShrinkIt (written entirely in assembly).  What I envision is
people using ShrinkIt at home on their Apple //s, but NuLib on UNIX
systems or other microcomputers.  This will facilitate transfers of
large compressed files, which can then be quickly unpacked on the
destination system (which will likely have greater computing power, or
a C compiler more efficient than APW).


System Dependencies
-------------------

When compiling this on a Sun 3/50, I noticed a problem: the byte
ordering for the //gs (65816) and the Sun (68020) are backward.  The
present version of NuLib will automatically determine the byte
ordering and treat data appropriately (although this may fail if the
data size definitions are wrong).

There are definitions for one byte, two byte, and four byte variable
types; my compiler uses char, short, and long.  If these are different
for your compiler, be sure to change the typedefs in "nudefs.h".  If
you don't have 8-bit bytes, though, it may not work (most machines
do).

Notes on UNIX implementation:
If this is being run under UNIX, you should #define UNIX in
"nudefs.h".  This will enable certain UNIX functions (like system()
calls and time routines), and disable others (like file types and
"binary mode" for file I/O).  If this is being run under BSD UNIX, you
should also #define BSD43 in "nudefs.h"; certain small differences
were unavoidable (strrchr() vs rindex()).  System V users should
#define SYSV.  See "nudefs.h" for examples.

Under APW, breaking down the full pathnames into relative pathnames is
easy (and should be under MS-DOS as well...).  Under UNIX ".." is an
actual directory link, so obtaining a fully expanded pathname with no
redundancies is difficult.  Thus, storing "dir1/../dir1/foo" would
appear as "dir1/foo" under APW but "dir1/../dir1/foo" under UNIX.
Care must be taken when extracting such files on non-UNIX systems;
they are best avoided entirely.  One nasty pitfall is using "~/foo"
under the C shell...

If the UNIX owner access permissions include write permission, the
file will be stored as unlocked.  If write permission is denied, it
will be stored as locked.  As far as dates are concerned, the
modification and access times will both be set to the modification
time.

Notes on //gs implementation:
There is very little //gs-specific code, except where absolutely
necessary (time routines, file type handling, etc).  GS/OS support is
absent (extended files are not handled yet, different file systems are
not recognized, and pseudo-mixed-case ProDOS filenames are stored as
upper case).  Some of the faults belong to APW...

Unfortunately, the startup code (2/start) provided with APW and the
methods of argument passing used by ECP-16 and ProSel-16 aren't
compatible.  This means that you can run NuLib from ECP or ProSel, but
you can't pass it any arguments (which pretty much defeats the
purpose...).

At any rate, the program is linked with the compression routines in
dynamic load segments, so that simple operations should run faster.
Additionally, it should be restartable from memory.  [actually, the
ZapLink version doesn't have dynamic segments...]

There is a bug (not my fault) when using the P option to print a file.
The file gets printed on one line for some reason...  If you redirect
the I/O to a file, everything comes out fine.  Weird.

APW shell-specific code has been avoided where possible.  Places where
I couldn't easily work around it include wildcard expansion and
ERROR().  It was deliberately used in several places to allow the user
to STOP() processing with Apple-period.

NOTE: NuLib was written to work with APW C, ** NOT **  Orca/C.
Attempting to compile it with Orca/C would be a major undertaking (as
several people who have tried and failed can attest to).

Notes on MS-DOS implementation:
The user interface may be slightly different from the UNIX and APW
versions, so that MS-DOS users will feel more at home when using the
program.  The only major deviation is in the handling of subdirectory
expansion.

A future version may select the ARC/ZOO output format as default.


Bugs / Glitches
---------------

UNIX lseek uses longs, which are usually four bytes.  Signed.  If an
archive is larger than 2 gigabytes, there may be a problem (cough).
This isn't a problem if the system has EFT.

Pathnames longer than the #defined maximum (1024 bytes) will not be
processed.  This is the limit on most machines, and is well in excess
of most people's sanity.  Pathnames with a null ('\0') in them should
generally be avoided.

Some partial pathname comparisons may fail because pathname separators
vary between operating systems.  File naming conventions can result in
collisions (ex: "foo+" and "foo~" are unpacked under ProDOS, where
both are translated to "foo.").  Also, you will probably need to be in
the same directory each time you U)pdate an archive, or else the
partial pathnames won't match (update requires an EXACT match), and
you'll end up with two different copies of the same file.

A maximum of 255 files may be added/deleted/whatever at at time.
Expanded subdirectories count.  This is an arbitrary number and is
easy to change if someone can convice me that you'd need to archive
more than 255 files at a time.

The same file may be added/extracted/whatever more than once if the
user enters multiple file selectors on the command line ("nulib av
foo.shk file1 file1 file1").  This means that U)pdate could insert the
same file twice, etc.

NuLib does not at present prevent an archive from being added to
itself.  This can have unfortunate consequences, especially in
conjunction with the M)ove option.

The ProDOS three-letter filetype names may or may not be the offical
Apple versions.  Some users may have problems with things like "$00"
because '$' is a csh metacharacter (job control).

Error output is informative but ugly (mostly debugging-type messages).
Error handling is somewhat fatal in most cases.

The code is fairly large, currently around 260K (around 7000 lines of
C).  It takes a while to compile it, and can be difficult if you only
have 1.25MB of RAM (kill your DAs and purge before you link!)


In the Works
------------

This is a wishlist of sorts.  Don't hold your breath:
- Allow CR <-> LF translation when ADDING files.
- Improve GS/OS handling.  Need to handle resource forks and file
system IDs. Would probably help if I had a GS/OS reference...
- Add to thread methods (insert ASCII messages, etc).
- Add different compression methods (LZH?).


Author Info
-----------

I'm currently an Associate Software Engineer for Amdahl Corp. in Santa
Clara, California.

e-mail: fadden@uts.amdahl.com (Andy McFadden)

No relation to the author of ShrinkIt and NuFX, Andy Nicholas.  This
was and continues to be my idea; throw all kudos and blame in my
general direction.  All code herein is mine, developed from the NuFX
Documentation Revisions three through five (the latter from Call
-A.P.P.L.E.), except as noted below.

Dynamic LZW-I (ShrinkIt) by Kent Dickey.  LZW-II compression (LZW-I
with table clears) designed by Andy Nicholas.  C implementation by
Frank Petroski and Kent Dickey (independently and simultaneously).

MS-DOS code by Robert B. Hess and Bruce Kahn.

The Binary II routines and the unSQueeze code were adapted from the C
source by Marcel J.E. Mol (usq.c based on usq2/sq3 by Don Elton).

UNIX compress code from COMPRESS v4.3 (see source for detailed author
information).

If you have suggestions or comments and can't send mail to my Internet
address, you can send US mail to:
Andy McFadden
1474 Saskatchewan Dr.
Sunnyvale, CA  94087

Andy Nicholas, the author of ShrinkIt and affiliated products
(ShrinkIt ][+, UnshrinkIt ][+, GS/ShrinkIt), may be contacted at:
     Internet    :  shrinkit@Apple.COM
     CompuServe  :  70771,2615
     GEnie and
      America OL :  shrinkit


Legal Stuff
-----------

(I'm not sure about all of these, but I'd like to cover all the
bases):

ShrinkIt is a trademark of L&L Productions, Inc (...or is it?).
ARC is owned by SeaWare.
Apple is a tradmark of Apple Computer, Inc.
MS-DOS is (probably) a trademark of Microsoft, Inc.
PKZIP is (probably) a trademark of PKWare.

I have not seen source code for ShrinkIt(tm) or for the compression
routines used within it, except as provided to me in C by Kent Dickey.

Disclaimers, borrowed from Dave Whitney and Andy Nicholas:

This program is FREEWARE; both source and binaries may be distributed
freely, but not sold.  If you wish to include NuLib in the
distribution of a commercial product, you will need to get my
permission beforehand.

Distribute as widely as possible, but please include this
documentation with the binaries and/or sources.

Copyright (C) 1989-1991 by Andy McFadden. All rights reserved.

I (Andy McFadden) MAKE NO WARRANTY ON THIS MANUAL OR SOFTWARE, EITHER
EXPRESS OR IMPLIED, WITH RESPECT TO QUALITY, MANUAL'S ACCURACY,
MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.

IN NO EVENT WILL I BE HELD RESPONSIBLE FOR DIRECT, INDIRECT, SPECIAL,
INCIDENTAL, OR CONSEQUENTIAL DAMAGES RESULTING FROM ANY DEFECT OF THE
SOFTWARE OR INACCURACY IN THE MANUAL.

In other words, I have tested the program to the best of my ability,
and I find that as distributed by me, it is safe for general use. It
isn't necessarily bug-free, and as a result, loss of data, however
unlikely, is entirely possible. Use at your own risk, but also for
your own enjoyment.