at [-m][-f file][-q
queuename] -t time_arg
at [-m][-f file][-q queuename]
at -r at_job_id ...
at -l -q queuename
The at utility shall read commands from standard input and group them together as an at-job, to be executed at a later time.
The at-job shall be executed in a separate invocation of the shell, running in a separate process group with no controlling terminal, except that the environment variables, current working directory, file creation mask, and other implementation-defined execution-time attributes in effect when the at utility is executed shall be retained and used when the at-job is executed.
When the at-job is submitted, the at_job_id and scheduled time shall be written to standard error. The at_job_id is an identifier that shall be a string consisting solely of alphanumeric characters and the period character. The at_job_id shall be assigned by the system when the job is scheduled such that it uniquely identifies a particular job.
User notification and the processing of the job's standard output and standard error are described under the -m option.
Users shall be permitted to use at if their name appears in the file /usr/lib/cron/at.allow. If that file does not exist, the file /usr/lib/cron/at.deny shall be checked to determine whether the user shall be denied access to at. If neither file exists, only a process with the appropriate privileges shall be allowed to submit a job. If only at.deny exists and is empty, global usage shall be permitted. The at.allow and at.deny files shall consist of one user name per line.
The at utility shall conform to the Base Definitions volume of IEEE Std 1003.1-2001, Section 12.2, Utility Syntax Guidelines.
The following options shall be supported:
If -m is not used, the job's standard output and standard error shall be provided to the user by means of mail, unless they are redirected elsewhere; if there is no such output to provide, the implementation need not notify the user of the job's completion.
Specify in which queue to schedule a job for submission. When used with the -l option, limit the search to that particular queue. By default, at-jobs shall be scheduled in queue a. In contrast, queue b shall be reserved for batch jobs; see batch. The meanings of all other queuenames are implementation-defined. If -q is specified along with either of the -t time_arg or timespec arguments, the results are unspecified.
The following operands shall be supported:
In the POSIX locale, the following describes the three parts of the time specification string. All of the values from the LC_TIME categories in the POSIX locale shall be recognized in a case-insensitive manner.
If no date is given, today shall be assumed if the given time is greater than the current time, and tomorrow shall be assumed if it is less. If the given month is less than the current month (and no year is given), next year shall be assumed.
at 2pm + 1 week at 2pm next week
The following grammar describes the precise format of timespec in the POSIX locale. The general conventions for this style of grammar are described in Grammar Conventions . This formal syntax shall take precedence over the preceding text syntax description. The longest possible token or delimiter shall be recognized at a given point. When used in a timespec, white space shall also delimit tokens.
%token hr24clock_hr_min %token hr24clock_hour /* An hr24clock_hr_min is a one, two, or four-digit number. A one-digit or two-digit number constitutes an hr24clock_hour. An hr24clock_hour may be any of the single digits [0,9], or may be double digits, ranging from [00,23]. If an hr24clock_hr_min is a four-digit number, the first two digits shall be a valid hr24clock_hour, while the last two represent the number of minutes, from [00,59]. */ %token wallclock_hr_min %token wallclock_hour /* A wallclock_hr_min is a one, two-digit, or four-digit number. A one-digit or two-digit number constitutes a wallclock_hour. A wallclock_hour may be any of the single digits [1,9], or may be double digits, ranging from [01,12]. If a wallclock_hr_min is a four-digit number, the first two digits shall be a valid wallclock_hour, while the last two represent the number of minutes, from [00,59]. */ %token minute /* A minute is a one or two-digit number whose value can be [0,9] or [00,59]. */ %token day_number /* A day_number is a number in the range appropriate for the particular month and year specified by month_name and year_number, respectively. If no year_number is given, the current year is assumed if the given date and time are later this year. If no year_number is given and the date and time have already occurred this year and the month is not the current month, next year is the assumed year. */ %token year_number /* A year_number is a four-digit number representing the year A.D., in which the at_job is to be run. */ %token inc_number /* The inc_number is the number of times the succeeding increment period is to be added to the specified date and time. */ %token timezone_name /* The name of an optional timezone suffix to the time field, in an implementation-defined format. */ %token month_name /* One of the values from the mon or abmon keywords in the LC_TIME locale category. */ %token day_of_week /* One of the values from the day or abday keywords in the LC_TIME locale category. */ %token am_pm /* One of the values from the am_pm keyword in the LC_TIME locale category. */ %start timespec %% timespec : time | time date | time increment | time date increment | nowspec ; nowspec : "now" | "now" increment ; time : hr24clock_hr_min | hr24clock_hr_min timezone_name | hr24clock_hour ":" minute | hr24clock_hour ":" minute timezone_name | wallclock_hr_min am_pm | wallclock_hr_min am_pm timezone_name | wallclock_hour ":" minute am_pm | wallclock_hour ":" minute am_pm timezone_name | "noon" | "midnight" ; date : month_name day_number | month_name day_number "," year_number | day_of_week | "today" | "tomorrow" ; increment : "+" inc_number inc_period | "next" inc_period ; inc_period : "minute" | "minutes" | "hour" | "hours" | "day" | "days" | "week" | "weeks" | "month" | "months" | "year" | "years" ;
The standard input shall be a text file consisting of commands acceptable to the shell command language described in Shell Command Language . The standard input shall only be used if no -f file option is specified.
See the STDIN section.
The text files /usr/lib/cron/at.allow and /usr/lib/cron/at.deny shall contain zero or more user names, one per line, of users who are, respectively, authorized or denied access to the at and batch utilities.
The following environment variables shall affect the execution of at:
When standard input is a terminal, prompts of unspecified format for each line of the user input described in the STDIN section may be written to standard output.
In the POSIX locale, the following shall be written to the standard output for each job when jobs are listed in response to the -l option:
"%s\t%s\n", at_job_id, <date>
where date shall be equivalent in format to the output of:
date +"%a %b %e %T %Y"
In the POSIX locale, the following shall be written to standard error when a job has been successfully submitted:
"job %s at %s\n", at_job_id, <date>
where date has the same format as that described in the STDOUT section. Neither this, nor warning messages concerning the selection of the command interpreter, shall be considered a diagnostic that changes the exit status.
The following exit values shall be returned:
The job shall not be scheduled, removed, or listed.
The format of the at command line shown here is guaranteed only for the POSIX locale. Other cultures may be supported with substantially different interfaces, although implementations are encouraged to provide comparable levels of functionality.
Since the commands run in a separate shell invocation, running in a separate process group with no controlling terminal, open file descriptors, traps, and priority inherited from the invoking environment are lost.
Some implementations do not allow substitution of different shells using SHELL . System V systems, for example, have used the login shell value for the user in /etc/passwd. To select reliably another command interpreter, the user must include it as part of the script, such as:
$ at 1800 myshell myscript EOT job ... at ... $
This sequence can be used at a terminal:
at -m 0730 tomorrow sort < file >outfile EOT
This sequence, which demonstrates redirecting standard error to a pipe, is useful in a command procedure (the sequence of output redirection specifications is significant):
at now + 1 hour <<! diff file1 file2 2>&1 >outfile | mailx mygroup !
To have a job reschedule itself, at can be invoked from within the at-job. For example, this daily processing script named my.daily runs every day (although crontab is a more appropriate vehicle for such work):
# my.daily runs every day daily processingat now tomorrow < my.daily
The spacing of the three portions of the POSIX locale timespec is quite flexible as long as there are no ambiguities. Examples of various times and operand presentation include:
at 0815am Jan 24 at 8 :15amjan24 at now "+ 1day" at 5 pm FRIday at '17 utc+ 30minutes'
The at utility reads from standard input the commands to be executed at a later time. It may be useful to redirect standard output and standard error within the specified commands.
The -t time option was added as a new capability to support an internationalized way of specifying a time for execution of the submitted job.
Early proposals added a "jobname" concept as a way of giving submitted jobs names that are meaningful to the user submitting them. The historical, system-specified at_job_id gives no indication of what the job is. Upon further reflection, it was decided that the benefit of this was not worth the change in historical interface. The at functionality is useful in simple environments, but in large or complex situations, the functionality provided by the Batch Services option is more suitable.
The -q option historically has been an undocumented option, used mainly by the batch utility.
The System V -m option was added to provide a method for informing users that an at-job had completed. Otherwise, users are only informed when output to standard error or standard output are not redirected.
The behavior of at <now> was changed in an early proposal from being unspecified to submitting a job for potentially immediate execution. Historical BSD at implementations support this. Historical System V implementations give an error in that case, but a change to the System V versions should have no backwards-compatibility ramifications.
On BSD-based systems, a -u user option has allowed those with appropriate privileges to access the work of other users. Since this is primarily a system administration feature and is not universally implemented, it has been omitted. Similarly, a specification for the output format for a user with appropriate privileges viewing the queues of other users has been omitted.
The -f file option from System V is used instead of the BSD method of using the last operand as the pathname. The BSD method is ambiguous-does:
at 1200 friday
mean the same thing if there is a file named friday in the current directory?
The at_job_id is composed of a limited character set in historical practice, and it is mandated here to invalidate systems that might try using characters that require shell quoting or that could not be easily parsed by shell scripts.
The at utility varies between System V and BSD systems in the way timezones are used. On System V systems, the TZ variable affects the at-job submission times and the times displayed for the user. On BSD systems, TZ is not taken into account. The BSD behavior is easily achieved with the current specification. If the user wishes to have the timezone default to that of the system, they merely need to issue the at command immediately following an unsetting or null assignment to TZ . For example:
TZ= at noon ...
gives the desired BSD result.
While the yacc-like grammar specified in the OPERANDS section is lexically unambiguous with respect to the digit strings, a lexical analyzer would probably be written to look for and return digit strings in those cases. The parser could then check whether the digit string returned is a valid day_number, year_number, and so on, based on the context.