We require contributors to sign our Contributor License Agreement and we don't have one on file for @xylar.

In order for us to review and merge your code, please e-sign the Contributor License Agreement PDF. We then need to manually verify your signature. We will ping the bot to refresh the PR status when we have confirmed your signature.

https://github.com/conda/conda/blob/main/conda/cli/main_init.py#L34

Needs a test.

@dholth, I would be happy to add a test but I would need some hand-holding as this is my first time attempting to contribute to conda.

We will probably be able to merge this. The argument is processed here in https://github.com/conda/conda/blob/main/conda/cli/main_init.py#L34 . There are enough negatives that it takes a while to convince ourselves that the new code is correct. It looks like the flag never worked properly.

        for_user = args.user
        if not (args.install and args.user and args.system):
            for_user = True
        if args.no_user:
            for_user = False

It looks like the flag never worked properly.

Yes, I belief that's the case. I did test this locally and it worked as expected.

(In the meantime, I'm working around this by just moving my .bashrc aside, running conda init, and moving my .bashrc back.)

Did you figure out why there is an args.user and and args.no_user

    setup_type_group.add_argument(
        "--user",
        action="store_true",
        help="Initialize conda for the current user (default).",
        default=NULL,
    )
    setup_type_group.add_argument(
        "--no-user",
        action="store_false",
        help="Don't initialize conda for the current user (default).",
        default=NULL,
    )

I don't believe there is a difference between having --user and not having any flag at all. So I think the --user flag should be removed. But since it isn't causing trouble, I didn't want that to complicate this particular PR.

Sorry for all the back and forth; related commits
https://github.com/conda/conda/blame/3218283763349416e384ee17f3b445a6187aa522/conda/cli/conda_argparse.py
https://github.com/conda/conda/blame/main/conda/cli/main_init.py#L34
9735f56
69e5d96
21b57d5

Here's what it "should" look like

    setup_type_group.add_argument(
        "--user",
        action="store_true",
        help="Initialize conda for the current user (default).",
        default=True,
        # suppress help since we want to emphasize --no-user ?
    )
    setup_type_group.add_argument(
        "--no-user",
        dest="user",
        action="store_false",
        help="Don't initialize conda for the current user.",
#        default=NULL, # set by --user flag
    )

and in main_init,

        for_user = args.user
        if not (args.install and args.user and args.system): # why should for_user be True if not all of install, user, system? were we trying to do something "default" if none were set?
            for_user = True
        if args.user == False:
            for_user = False

Here's what it "should" look like

    setup_type_group.add_argument(
        "--user",
        action="store_true",
        help="Initialize conda for the current user (default).",
        default=True,
...

Okay, I'm willing to give this default=True a try. I'll not have a chance to thoroughly test until tomorrow (it's 10 pm where I live).

Thanks! We'll get this done carefully.

Also, happy to squash my commits to keep things cleaner, I just don't want to cause any undue confusion.

Okay, I couldn't let this go and I did some more testing. The logic works up to this line:

args.user = False # yay!
args.system = NULL
args.install = NULL

It seems that currently the only way to prevent my .bashrc from being altered is to run:

conda init --no-user --install --system

I can't imagine this is the intended behavior.

Looking back at the commit history, I think:

My sense at this point is that this line just doesn't make sense anymore at all and should be removed. I'm going to do that (and include this long explanation).

With my latest change, I have verified that conda init and conda init --user both produce:

...
no change     /home/xylar/Desktop/mambaforge/etc/profile.d/conda.csh
modified      /home/xylar/.bashrc

but that conda init --no-user does not:

no change     /home/xylar/Desktop/mambaforge/etc/profile.d/conda.csh
No action taken.

@dholth, please give this another look when you have time.

https://docs.conda.io/projects/conda/en/latest/commands/init.html?highlight=Conda%20init

It makes intuitive sense that --system could disable --user. We just have to replace the old logic.

Looks good but this line must be a mistake? We set for_user = True if install is false, or user is false, or system is false

Just saw this comment

It makes intuitive sense that --system could disable --user. We just have to replace the old logic.

Okay, I agree. Let's make this work the way it's supposed to!

So if someone specifies both --system and --user, shouldn't there just be an error?

And what is --install supposed to do (I'll try to figure it out)?

@dholth, just wondering about the process here. Is there anything more I need to do to see this get merged at some point?

Thanks, I've tagged this appropriately. We will merge it within a couple of weeks +- holidays

Updated to fix the tests/conda_env/test_create.py::IntegrationTests::test_create_update_remote_env_file test failures. Should pass now.

See: https://github.com/conda/conda/actions/runs/3414753536/jobs/6144233243#step:7:709
Fix: #12024

🤞🏻