(More) complete support for depends_on

[rtoohil]

My compose file has a more complex use of the depends_on functionality than Container-Compose used.

I will admit that Swift isn't my best language, so I worked with Claude Code to build this PR. This adds support for various depends_on syntaxes such as:

1. Simple string (no colon after service name):

  depends_on: mysql

  2. Simple array:

  depends_on:
    - mysql
    - redis

  3. Dictionary without conditions (empty/null values):

  depends_on:
    mysql:
    redis:

  4. Dictionary with conditions (full format):

  depends_on:
    mysql:
      condition: service_healthy
    redis:
      condition: service_started

It doesn't handle the health check well yet, but we can expand upon that. This handled my more complex depends_on case.

[Mcrich23]

Thank you. I'm curious about your thoughts on this general idea given the lack of support for the features described in your compose file.

[rtoohil]

Honestly, part of the idea was simply to reduce the friction for someone getting started, i.e. I have a working compose file that I use with docker compose. Rather than forcing users to rewrite, can handle the array/dictionary case to allow someone's existing file to get things going. If desired, for things like condition, could log that those are not yet implemented/supported.

I can see the argument for not supporting the condition case at all, but I still think the friction reduction is worth it (one singular opinion here)

[Mcrich23]

This is true. I will do a manual review of your PR when I have a chance. One way that would be incredibly helpful to contribute if you would like would be to document the current status of compose yaml feature support.

Feel free not to if you don't want to or don't have the capacity to.

[rtoohil]

I should be able to take a crack at that. No rush on my PRs — I just happened to have some time and thought I would try to contribute.

…

On Dec 1, 2025, at 12:43 PM, Morris Richman ***@***.***> wrote: Mcrich23 left a comment (Mcrich23/Container-Compose#35) This is true. I will do a manual review of your PR when I have a chance. One way that would be incredibly helpful to contribute if you would like would be to document the current status of compose yaml feature support. Feel free not to if you don't want to or don't have the capacity to. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[Mcrich23]

Thank you. Can you please add tests for this? I want to ensure in the future that this kind of compose file will continue to build.

[rtoohil]

Just as an update — working on the tests, but adding the tests highlighted one of the original tests that was passing was actually failing. Seeing if I can get that working as part of this PR

…

On Dec 1, 2025, at 6:16 PM, Morris Richman ***@***.***> wrote: Mcrich23 left a comment (Mcrich23/Container-Compose#35) Thank you. Can you please add tests for this? I want to ensure in the future that this kind of compose file will continue to build. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[Mcrich23]

Thank you. It appears all of the tests are passing on main which would mean that it is an issue in your pr.

[rtoohil]

They are all passing on main, which is when I started to investigate, but the test "Test compose with complex dependency chain” returns success even though a couple of the containers never start due to errors. When updating the depends_on logic, we catch this error now, so the test fails. I’m partially through debugging why the containers are failing (one was the wrong argument passed into it — which is now working, but still hunting it down). If you see different behavior (i.e. the containers for app and wordpress do start properly), let me know. But I think I’ll make progress on fixing that issue in the next day or so.

…

On Dec 3, 2025, at 12:55 PM, Morris Richman ***@***.***> wrote: Mcrich23 left a comment (Mcrich23/Container-Compose#35) Thank you. It appears all of the tests are passing on main which would mean that it is an issue in your pr. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[rtoohil]

Sorry for the long delay (real life). This has the same tests passing as on main (the Real World tests fail on main on my machine), but I reverted the change that broke the other test. That test is now passing.

I do think that there's some command handling that may need to change to get these other tests working, but that can be a separate PR.

[Mcrich23]

No problem! Let me create a new branch for this feature when I'm back at my computer and we can merge this into there instead. Then we can make a new pr for tests.

[Mcrich23]

Havent tried it, but it lgtm. Lets get this merged into the new feature branch i made and then do more testing before merging to main.