Git, Composer, Node and NPM not installed, then a permission error

[XorOwl]

after the recent update, my install went weird on me, with the database being inaccessible, so i started a new folder to start from scratch and still had problems.
I shut down the existing instance and purge all disks and cache in docker
I fresh clone the repo and submodules, copy my .env and pathfinder.ini
i cut out the traefic elements and add a port that i have nginx reverse proxying to
then i run "docker network create web && docker-compose up -d" and open the web browser
The browser displays an infinite loading screen
viewing the log while refreshing on setup shows the following:

pathfinder | [27-Jan-2024 18:56:18] WARNING: [pool www] child 22 said into stderr: "sh: git: not found"
pathfinder | [27-Jan-2024 18:56:18] WARNING: [pool www] child 22 said into stderr: "sh: composer: not found"
pathfinder | [27-Jan-2024 18:56:18] WARNING: [pool www] child 22 said into stderr: "sh: node: not found"
pathfinder | [27-Jan-2024 18:56:18] WARNING: [pool www] child 22 said into stderr: "sh: npm: not found"

bashing into the pf instance and running "apk add git composer nodejs npm --update" fixes this, but it must be done any time the disk is rebuilt.

After doing so i get

pathfinder | [27-Jan-2024 19:04:00] WARNING: [pool www] child 91 said into stderr: "Cannot create cache directory /.composer/cache/repo/https---packagist.org/, or directory is not writable. Proceeding without cache"
pathfinder | [27-Jan-2024 19:04:00] WARNING: [pool www] child 91 said into stderr: "Cannot create cache directory /.composer/cache/files/, or directory is not writable. Proceeding without cache"

before i get MASSIVE blocks of whitespace echo'd into stderr

in the middle of the whitespace is an error about "Unknown package has no name defined". It's rather long so this is included as an attachment pflog.txt

Is there maybe an incorrect image being used that needs to be better specified in the docker-compose file?

pflog.txt

[XorOwl]

it seems a lot of what i listed above was red herrings.
Somewhere, a timeout was too short, truncating some pages that were trying to load, causing all kinds of weird behavior.
Using the internal ip address, and configuring the server to use that, all the initial setup steps work.