From 70accc0e170761098cd6fa4e0ec325c3d7ed597f Mon Sep 17 00:00:00 2001
From: WXTIM <26465611+wxtim@users.noreply.github.com>
Date: Fri, 28 Jun 2024 12:02:53 +0100
Subject: [PATCH] added simplest example

---
 simplest-useful/bin/get_data.sh     | 20 +++++++++++
 simplest-useful/bin/process_data.sh |  8 +++++
 simplest-useful/flow.cylc           | 56 +++++++++++++++++++++++++++++
 simplest-useful/simple_script.sh    |  8 +++++
 4 files changed, 92 insertions(+)
 create mode 100755 simplest-useful/bin/get_data.sh
 create mode 100755 simplest-useful/bin/process_data.sh
 create mode 100644 simplest-useful/flow.cylc
 create mode 100755 simplest-useful/simple_script.sh

diff --git a/simplest-useful/bin/get_data.sh b/simplest-useful/bin/get_data.sh
new file mode 100755
index 0000000..ecf6822
--- /dev/null
+++ b/simplest-useful/bin/get_data.sh
@@ -0,0 +1,20 @@
+#!/bin/bash
+
+HEAD="[MY SUPERCOMPUTER] "
+
+echo "${HEAD}Waiting for resources"
+sleep 7
+echo "${HEAD}"
+
+if [[ -f output ]]; then
+    rm output
+fi
+
+if [[ $((RANDOM % 4)) == 0 ]]; then
+    echo "Data Retrieval Succeeded"
+    touch output
+    exit 0
+else
+    echo "Data Retrieval Failed"
+    exit 1
+fi
diff --git a/simplest-useful/bin/process_data.sh b/simplest-useful/bin/process_data.sh
new file mode 100755
index 0000000..705c89b
--- /dev/null
+++ b/simplest-useful/bin/process_data.sh
@@ -0,0 +1,8 @@
+#!/bin/bash
+
+cylc message -- "what is the shortest possible timespan for a pretend hpc job?"
+if [[ ! -f output ]]; then
+    echo 'ERROR - NO INPUT DATA!'
+    exit 1
+fi
+sleep 14
diff --git a/simplest-useful/flow.cylc b/simplest-useful/flow.cylc
new file mode 100644
index 0000000..f6884fb
--- /dev/null
+++ b/simplest-useful/flow.cylc
@@ -0,0 +1,56 @@
+[meta]
+    title = Pretty much the simplest use case.
+    description = """
+    ## What is the smallest workflow worth converting to Cylc?
+
+    Consider the case where you want to get some data and
+    do some number crunching:
+
+    ```bash
+        #!bin/bash
+        #@supercomputer --time 300
+        #@supercomputer --memory LOTS
+        #@supercomputer --CPU MANY
+
+        get_data.sh
+
+        process_data.sh
+    ```
+
+    Where `get_data.sh` is a data retrieval programme:
+
+    * High IO & Slow
+    * Low Memory and CPU requirements
+    * Might be flaky?
+
+    And `process_data.sh`
+
+    * High memory and CPU
+
+    ## Why use Cylc
+
+    #. Don't ask for resources until pre-reqs are done
+    #. Retry flaky steps
+    #. Using `execution time limit` means that Cylc knows that
+       a task has taken too long even if communications are
+       disrupted.
+    """
+
+[scheduling]
+    [[graph]]
+        R1 = get_data => process_data
+
+[runtime]
+    [[get_data]]
+        script = get_data.sh
+        platform = any_old_server
+        execution retry delays = 4*PT15M, PT1D
+
+    [[process_data]]
+        script = process_data.sh
+        platform = supercomputer
+        # Cylc converts to #@supercomputer --time 300
+        execution time limit = PT5M
+        [[[directives]]]
+            --memory LOTS
+            --CPU MANY
diff --git a/simplest-useful/simple_script.sh b/simplest-useful/simple_script.sh
new file mode 100755
index 0000000..6b5e7b1
--- /dev/null
+++ b/simplest-useful/simple_script.sh
@@ -0,0 +1,8 @@
+#!/bin/bash
+#@supercomputer --time 300
+#@supercomputer --memory LOTS
+#@supercomputer --CPU MANY
+
+./bin/get_data.sh
+
+./bin/process_data.sh