diff options
author | Joe Hunkeler <jhunkeler@gmail.com> | 2015-08-11 16:51:37 -0400 |
---|---|---|
committer | Joe Hunkeler <jhunkeler@gmail.com> | 2015-08-11 16:51:37 -0400 |
commit | 40e5a5811c6ffce9b0974e93cdd927cbcf60c157 (patch) | |
tree | 4464880c571602d54f6ae114729bf62a89518057 /pkg/language/doc/prcache.hlp | |
download | iraf-osx-40e5a5811c6ffce9b0974e93cdd927cbcf60c157.tar.gz |
Repatch (from linux) of OSX IRAF
Diffstat (limited to 'pkg/language/doc/prcache.hlp')
-rw-r--r-- | pkg/language/doc/prcache.hlp | 94 |
1 files changed, 94 insertions, 0 deletions
diff --git a/pkg/language/doc/prcache.hlp b/pkg/language/doc/prcache.hlp new file mode 100644 index 00000000..62398f16 --- /dev/null +++ b/pkg/language/doc/prcache.hlp @@ -0,0 +1,94 @@ +.help prcache Feb86 language +.ih +NAME +prcache -- cache a subprocess +.ih +USAGE +prcache task [task ...] +.ih +PARAMETERS +.ls task +The name of a compiled IRAF task (not the filename of an executable file). +.le +.ih +DESCRIPTION +The CL maintains a small cache to store executable images. When the +user invokes a task which calls an executable the cache is searched +for the image before any attempt to load the executable image from +disk. After completion of the task the CL retains the executable image +in the process cache, until the space is needed by some other +executable. Thus if a few commands are being executed frequently, +the overhead of loading the image into core from disk is bypassed, +which can result in a significant improvement in the response of the CL. + +By default, when the cache is full and a new executable must be run, +the CL searches for the slot containing the task which has remained +dormant the longest and replaces it with the new task. + +The \fIprcache\fR command gives the user some control over this process. +Using it without any arguments shows the tasks which are currently +stored in the process cache. For each slot one gets a line like the +following. + + [07] lyra!17763(4563X) H bin$x_images.e + +Here, 07 is the process slot number as required by \fIflprcache\fR to +disconnect the process. The name "lyra" is the name of the node in the +local network on which the process is executing; this is not normally +the local node. In the example, 17763 (hex 4563X) is the process number +(pid) of the executable. H indicates that the task is hibernating, +i.e. the task was waiting in the cache to be invoked. R would show that +the task was running. An L appended to either of these would show that the task +had been locked into the cache by a previous prcache command. +The last element on the line is the file name of the executable +file which was loaded when the task was first invoked. + +If one or more task names are given as arguments, those tasks are locked +into the cache, and will not be replaced by the CL without specific user +intervention. If these tasks are not already in the cache, the +corresponding executables are started, and the tasks are loaded into the cache. + +Note that the `process cache' described here, and the `parameter cache' +described in the `cache' command are entirely distinct, and a given +task may be found in either, both, or neither of the two caches. + +Also note that only executable images reside in the process cache. +Thus, for example, if the \fInews\fR command is executed, it does not +appear in the process cache, but the executable +`system$x_system.e' does, because \fInews\fR calls \fIpage\fR, +which is one of the many entries into the system executable. + +Locked process cache slots may only be freed with the \fIflprcache\fR command. +.ih +EXAMPLES +1. Flush the system process and lock it back into the cache. + +.nf + cl> flpr dir + cl> prc dir +.fi + +2. Print the current contents of the process cache. + +.nf + cl> prc + [10] lyra!17764(4564X) H bin$x_plot.e + [07] lyra!17763(4563X) H bin$x_images.e + [04] lyra!17455(442FX) HL bin$x_system.e + 0 +.fi + +3. Flush all processes which are not locked into the cache. This may be +necessary after aborting a task to initialize (by re-executing) the +associated process, which may not have recovered completely from the +abort. + + cl> flpr +.ih +BUGS +The user is responsible for making sure that he does not lock all +the slots in the cache. +.ih +SEE ALSO +flprcache +.endhelp |