dp-redo 3.0.1

A python variant of the redo build system, with which You Can (Not) Redo.

dp-redo

A python variant of the redo build system, with which You Can (Not) Redo.

Example

Let's say we have a build process where we read test.c and generate test.o:

import os, sys
source_tree = os.path.dirname(os.path.abspath(sys.argv[0]))

def test():
    source = os.path.join(source_tree, "test.c")

    # The simplest compiler 😁
    os.system("cat {} > {}".format(source, "test.o"))

if __name__ == "__main__":
    test()

test will be run every time no matter whether test.c gets an update or not, which wastes much time. Now let's add some magic:

from dp_redo import *
import os, sys

source_tree = os.path.dirname(os.path.abspath(sys.argv[0]))

@do("test.o")
def test(target_name, target_base_name, output_path):
    redo_ifchange("test.c")
    source = os.path.join(source_tree, "test.c")
    os.system("cat {} > {}".format(source, output_path))

if __name__ == "__main__":
    redo_ifchange(test)

If you run it multiple times:

PS C:\Users\lishengq\source\repos\dp-redo\build> python ..\test2.py
Redoing target: test
PS C:\Users\lishengq\source\repos\dp-redo\build> python ..\test2.py
Skipping target test: it's up to date.

Fantastic, isn't it? If you changed test.c, modified the test method itself, or deleted test.o, test will be executed again:

PS C:\Users\lishengq\source\repos\dp-redo\build> python ..\test2.py
Redoing target: test. Reason: It's python code has changed
PS C:\Users\lishengq\source\repos\dp-redo\build> python ..\test2.py
Redoing target: test. Reason: The target file doesn't exist
PS C:\Users\lishengq\source\repos\dp-redo\build> python ..\test2.py
Redoing target: test. Reason: test.c has been modified
PS C:\Users\lishengq\source\repos\dp-redo\build> 

---

A target method can also depend on another method, which is rather simple:

from dp_redo import *
import os, sys

source_tree = os.path.dirname(os.path.abspath(sys.argv[0]))

@do("test2.o")
def test2(target_name, target_base_name, output_path):
    os.system("echo test2 > " + output_path)

@do("test.o")
def test(target_name, target_base_name, output_path):
    redo_ifchange(test2, "test.c")
    source = os.path.join(source_tree, "test.c")
    os.system("cat {} > {}".format(source, output_path))

if __name__ == "__main__":
    redo_ifchange(test)

---

What makes redo more interesting is that, you can call redo_ifchange to add dependencies at any time, even after your compilation:

from dp_redo import *
import os, sys, re

source_tree = os.path.dirname(os.path.abspath(sys.argv[0]))

@do("test2.o")
def test2(target_name, target_base_name, output_path):
    print("In test2")
    os.system("echo test2 > " + output_path)

@do("test.o")
def test(target_name, target_base_name, output_path):
    redo_ifchange(test2)
    source = os.path.join(source_tree, "test.c")
    os.system("gcc -M -MF test.c.dep -o {} {}".format(output_path, source))

    # Dependent headers given by gcc
    deps = open('test.c.dep', 'r').read().split(": ")[1].strip().split("\\\n")
    redo_ifchange(*deps)

if __name__ == "__main__":
    redo_ifchange(test)

Now when any headers are changed, the build process will be run again.