Try using a CompiledComponentsPythonRecipe - you'll need to do something like this to get the argon C components to build.

I tried to look into what is a CompiledComponentsPythonRecipe but this is what it says in the documentation:

Is this link relevant to what you think I should look for : (https://askpythonquestions.com/2020/12/13/kivy-python-for-android-how-to-create-custom-recipes/)

You may (also) need to set call_hostpython_via_targetpython = False; e.g.

From what I understand it's still not able to compile the c?

From what I understand it's still not able to compile the c?

It can't find the argon.h file do I need to include the c code somehow?

It can't find the argon.h file do I need to include the c code somehow?

Looks like it uses a git submodule for the extras/libargon2 directory, which is not included in the .zip.

Try using url = 'git+https://github.com/hynek/argon2-cffi@{version}'. I think that will include submodules.

It can't find the argon.h file do I need to include the c code somehow?

Looks like it uses a git submodule for the extras/libargon2 directory, which is not included in the .zip.

shall I make my own zip with the other repo included inside this one from: https://github.com/p-h-c/phc-winner-argon2/tree/62358ba2123abd17fccf2a108a301d4b52c01a7c

Try using url = 'git+https://github.com/hynek/argon2-cffi@{version}'.

okay, sorry I typed my thing before seeing your comment

Try using url = 'git+https://github.com/hynek/argon2-cffi@{version}'. I think that will include submodules.

Actually, I think that should just be url = 'git+https://github.com/hynek/argon2-cffi'.

Normally, you should include {version} in the url, but the @{version} syntax for git urls is for pip, whereas p4a seems to use the recipe's .version instead (and check out that tag/branch).

Okay, I don't quite understand what the issue is now? It seemed like it almost passed it this time though

I think the problem is that p4a calls setup_ext but not setup_clib.
Try adding build_cmd = 'build' to the recipe. That should call both.

You may (also) want to try building the recipe on your own computer instead.
With buildozer, you can use e.g.

p4a.branch = develop
p4a.local_recipes = ./my-p4a-recipes


I don't understand how the p4a.local_recipes works, this is what I have so far:

If I am using my own fork, then why does it need the local_recipes thing?

Incase you are wondering if I have done a build with those settings above in me .spec file, here is the output
build_output.txt

If I am using my own fork, then why does it need the local_recipes thing?

In that case it doesn't :)
I'm using it so that I don't need to use a fork (and keep it up to date) just to have an extra recipe.

Incase you are wondering if I have done a build with those settings above in me .spec file, here is the output
build_output.txt

Oh okay, so if I use a local recipe do I still need to specify the package in the requirement field? Also what do you think of that build output I uploaded, is that some mistake on my part?

This recipe seems to work:

You may run into #2400 when testing that locally.

i'll do a rebase and squash the commits

I just added the change in the url

yeah, so as I thought, the libargon2 is missing hence we used that url from before, so this time I am gonna make the changes, squash the commits, and force push.

I am getting a lot of merge conflicts even though I followed the steps to a git rebase properly

Thanks for giving it a try
Weird for the rebase. I can't think of what would have made conflicts, but that's fine we'll try to deal without.
FYI what I would have tried is:

I'll try to at least squash from the GitHub interface at least because it allows to

Thanks for giving it a try
Weird for the rebase. I can't think of what would have made conflicts, but that's fine we'll try to deal without.
FYI what I would have tried is:

git fetch --all --prune
# squashing
git reset --soft HEAD~17
git commit --amend
# rebasing
git rebase upstream/develop # given the remote is called upstream

I'll try to at least squash from the GitHub interface at least because it allows to
I'm sorry, I was trying to post the comment and closed the pr by mistake, the issue I think is that I mixed up the commits, I did a few directly from github instead of the command line

okay tests passed, no more commits now. Just gotta squash the commits from before.

@AndreMiras I tried the git commands you suggested the git log on my local repo gives the 17 commits even after running those commands

No problem, thanks again for trying

git fetch --all --prune
# squashing
git reset --soft HEAD~17
git commit --amend
# rebasing
git rebase upstream/develop # given the remote is called upstream

Using an interactive rebase to squash always works for me:

yeah, so as I thought, the libargon2 is missing hence we used that url from before, so this time I am gonna make the changes, squash the commits, and force push.

An alternative solution would be to add a separate recipe for libargon2 and depend on that.
This might be a good idea anyway, since that can then be updated separately (and shared by other recipes).

@obfusk just give me a few days, I am in the middle of my semester exams, I will get back to this and clean it up as well as try and do this the right way. :) Thanks a lot for your help.

just give me a few days, I am in the middle of my semester exams

np. There's no hurry :) Good luck!

I will get back to this and clean it up

The current version is squashed and merged. So it's already fine :)

as well as try and do this the right way.

There's nothing "wrong" now. Using a separate libargon2 recipe would just be a nice improvement IMHO. And should be a separate PR.