Following the command git submodule update I have the output
Cloning into '/home/paulssd2/DT-Mast-15Jan20/darktable-code/src/tests/integration'...
This step has taken a long time (minutes) and downloaded 455Mb of what I’m guessing is the integration test pack!
I presume this is a mistake; or can it be me?
Out of interest, are the expected images checked by eye or is it automated?
You need to read the darktable-dev mailing list. This is intentional.
Automated, based on DeltaE on image.
@Pascal_Obry sent this to both the user and dev lists:
If you don’t care about the integration tests and want to avoid
downloading the tests you can do:$ git config submodule.src/test/integration.update none
And when you update, it’s recommended not to use the --recurse-submodules option to git pull.
I’m about to test this myself. 
And PLEASE USE UPDATED COMMAND, I missed the s in the first, the command is:
$ git config submodule.src/tests/integration.update none
1 Like
Yes, thanks. An easy way to test whether it’s effective: execute git submodule update. If its output includes Skipping submodule 'src/tests/integration', all should be OK.
1 Like
Thanks folks, I’ll do this next time I build.
Am I right to say the sequence (or a sequence) is -
git clone …
git checkout master
git config submodule.src/tests/integration.update none
git submodule init
git submodule update
1 Like
Once you’ve cloned the repository, you don’t need to do it again. Same for checking out the master branch and for submodule init.
git clean -d -f -x # gets rid of any local changes, including the previous 'build' directory, so you start with a clean slate
git submodule update # updates submodules
git pull --rebase # gets the latest changes for your current branch
./build.sh ...
Since I have an old computer, I start my script with the following commands, reducing the CPU and I/O priorities of the build process:
renice -n 19 -p $$
ionice -c 3 -p $$
You only need to tell git once that you don’t want them.