home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
C/C++ Users Group Library 1996 July
/
C-C++ Users Group Library July 1996.iso
/
vol_100
/
123_01
/
ar1.doc
< prev
next >
Wrap
Text File
|
1985-03-09
|
8KB
|
284 lines
.bp 1
.in 0
.he 'AR1 (1)'03/10/83'AR1 (1)'
.fo ''-#-'
.fi
.in 7
.ti -7
NAME
.br
ar1 - archive file maintainer
.sp 1
.ti -7
SYNOPSIS
.br
ar1 [-](d|p|t|u|x)[v] archive [files]
.sp 1
.ti -7
DESCRIPTION
.br
AR1 collects sets of arbitrary files into one big file and
maintains that file as an 'archive'.
Files can be extracted from the archive, new ones can be added,
old ones can be deleted or replaced by updated versions, and
data about the contents can be listed.
If a minus sign ('-') is given as a file name, further file names
are read from the standard input, one file name per line.
Files that are to be added to an archive must exist as
files with the name given. Files that are extracted from
an archive will be put onto files with the name given.
Files that are added to archives can, of course, be archive files
themselves. There is no (theoretical) limit to the number
of files that can be nested this way.
Thus AR1 provides the utility necessary to maintain tree-structured
file directories.
AR1 is invoked by the command line
.ti +10
AR1 command archname [optional filenames]
where 'command' is any one of 'dptux', optionally concatenated
with 'v', specifying what operation to perform on the
archive file named 'archname'.
The possible commands are:
.br
.in +10
u - Update named archive by replacing existing files or adding new
ones at end.
If the 'v' option is used, file names will be printed on the
standard output as files are written to the new archived file.
x - Extract named files from archive. Put onto file of the
same name.
If the 'v' option is added, file names will be printed on the
standard output as files are extracted.
d - Delete named files from archive.
If the 'v' option is used, file names will be printed on
the standard output as they are deleted from the archive.
p - Print named files on standard output.
Using the 'v' option will cause the file name to precede the file.
t - Print table of archive contents.
Normally, the table will contain only the file name.
If the 'v' option is used, the table will also contain
the file's length, type, and date and time of last change.
v - Verbose. This command may be concatenated to any of the above
commands,
and will cause the archiver to print additional information,
generally file names, on the standard output.
Its specific action for each command has already been described.
.br
.in -10
The optional filenames in the command line specify individual
files that may participate in the action.
If no files are named, the action is done on ALL files in the archive,
but if any files are explicitly named, they are the ONLY ones that
take part in the action.
(The 'd' command is an exception--files may be deleted only by
specifying their names.)
.sp 1
.ti -7
FILES
.br
Files of the form 'temp.$n$' are created and subsequently deleted for each run.
.sp 1
.ti -7
SEE ALSO
.br
The Unix commands 'ar' and 'ls' in the Unix manual
.br
The RATFOR AR2 tool.
.sp 1
.ti -7
DIAGNOSTICS
.br
archive not in proper format
.br
.in +10
The basic problem is that archive didn't find a header
line where one was expected.
Typical reasons include misspelling the file name,
using an existing file (not in archive format)
on a creation run, and
referencing an archive file that has been modified
directly (say with the editor).
.br
.in -10
archive integrity in doubt - missing trailer
.br
.in +10
Each file in an archive is terminated by a special line called a
"trailer," which must be present.
The message is caused by the lack of a trailer line when one was
expected.
.br
.in -10
delete by name only
.br
.in +10
For user protection, files are allowed to be deleted from an archive
only by specifying each file name.
.br
.in -10
duplicate file name
.br
.in +10
A file was listed more than once when calling the archiver
.br
.in -10
fatal errors-archive not altered
.br
.in +10
This message is generated whenever one or more of the other
errors have been detected. An archive is never altered unless
EVERYTHING has run properly.
.br
.in -10
too many file names
.br
.in +10
At the present the user may call the archiver with no more than
25 files at a time.
.br
.in -10
usage: ar [-](dptux)[v] archive [files]
.br
.in +10
The command line passed to the archiver is in error.
Possibly the command is wrong or the archive file name
has not been given.
.br
.in -10
can't add 'filename'
.br
.in +10
The file specified by 'filename' doesn't exist or can't be
opened (e. g. is locked).
.br
.in -10
can't create 'filename'
.br
.in +10
The archiver could not generate a local file by the name
of 'filename'. Probably the archiver's internal file
buffer space has been exceeded.
.br
.in -10
'filename' not in archive
.br
.in +10
The archiver could not locate the file specified by 'filename' in
the archived file.
.br
.in -10
.sp 1
.ti -7
AUTHORS
.br
Original code from Kernighan and Plauger's 'Software Tools',
with rewrites by Allen Akin (Georgia Institute of Technology)
and minor changes suggested by David Hanson (University
of Arizona).
Transliterated from RATFOR to BDS C by Oscar Goldman.
Conversion to BDS C library by Edward K. Ream
.sp 1
.ti -7
BUGS/DEFICIENCIES
.br
AR1 is quite slow with floppies, but tolerable with hard disks
and especially with memory disks.
One reason for its slow speed is that it ALWAYS reads the entire archive
so that it can check for a proper end to the archive.
Omitting this check would increase its speed.
A different approach (taken by AR2) is to have the index either as
a separate file, or as part of the archive file occupying the first
several records.
The index would consists of the name, some
statistical information and pointers into the archive file indicating
the starting and ending records as well as byte addresses within these
records.
This would certainly speed up access and would surely increase
the usefulness of the routine.
Another problem with AR1 arises because all reads and writes are
by line. For example, a constituent of an archive must end with a
NEWLINE, otherwise the trailer is messed up and the 'archive integrity'
is in doubt.
Line access also prevents archiving binary files.
A different approach to archiving CP/M files is clearly called for.
However, it is still not clear how to tell ascii files from other files.
When the update and print commands are used, the files are
updated or printed
in the order they appear on the
archived file, NOT the order listed on
the command line.
The Unix archiver allows files to be positioned in the
archive, rather than simply added at the end as AR does.
This is done by adding the following commands:
.br
.in +10
m - Move specified files to end of archive
ma posname - Move specified files to position after file 'posname'
mb posname - Move specified files to position before file 'posname'
r - Replace specified files and place at end of archive
ra posname - Replace files and place after file 'posname'
rb posname - Replace files and place before file 'posname'
.br
.in -10
There are some discrepancies between the Unix version of AR and
this version. Unix uses 'r'--replace instead of 'u'--update.
Unix also requires the user to specify an additional command 'n'
when creating a new archive.
.sp 1
.ti -7
NOTES
.br
1. The standard wildcard matching supplied in the wildexp package can NOT
be used, since wildexp expands file names by looking at files on the
DISK.
In general, the names of files on the DISK have nothing to do with
the names of the files in the ARCHIVE.
If you feel the need for wildcard matching, make the change in the
file_arg() routine.
2. AR1 can be used to unpack the files from the Software Tools project.
3. Link AR1 as follows:
clink ar1 -f dio ar1/util
4. At present, the getnow() routine in ar1/util.bds just prompts the
console for a time stamp.
The files tod.bds and date/ssi.bds show how to replace getnow()
with more clever code if you have clock or