Automated project completion tracking

[razorblade23]

As we have GH PAT login and lesson progress tracking, it would be nice to implement repository templates for projects on the platform.

The plan is for students to clone template which contains GitHub Actions pipeline so it automaticly runs tests on the repository and pushes progress to platform

[JeelDobariya38]

if setup repository template will it be all exercises in one template or a seperate exercise per repository.. if its like that then don't think the students & the python ledger will get filled with repository's...

I have a idea, can't we make it such that we make main responsibility in python ledger and make setup like user will download a cli application pyledger login to login to account using access token or something. then we need to pyledger test 01-hello-world & then pyledger submit 01-hello-world they will fet h the python - pytest/unittest test file from github repostory like ThePythonLedger/Excersise-Test and download necessary test file in .pyledger-temp and then run test files simple...

idk wether that is possible or not

that way github action machine will be saved, also your account will saved from cluttering but user progress can't be saved as we don't have backend or db server for ledger..

[razorblade23]

if setup repository template will it be all exercises in one template or a seperate exercise per repository.. if its like that then don't think the students & the python ledger will get filled with repository's...

Currently, projects are in curriculum repository and i think they will stay there.

I have a idea, can't we make it such that we make main responsibility in python ledger and make setup like user will download a cli application pyledger login to login to account using access token or something. then we need to pyledger test 01-hello-world & then pyledger submit 01-hello-world they will fet h the python - pytest/unittest test file from github repostory like ThePythonLedger/Excersise-Test and download necessary test file in .pyledger-temp and then run test files simple...

idk wether that is possible or not

Yes, its definetly possible and this was also my idea too. As for running the tests, it can be done on GH Actions or i can spin up a quick WASM based python interpreter, run the tests and report back to user/student - all thrue the CLI app. Way faster then GH Actions and local - no sending data anywhere.

This is plannned for actuall projects done on the paltform, not exercises - these are meant just as a more practice for learners, nothing more.

Projects are what comes after the chapter/chapters is what i hope to automate.

Exercises are standalone thing and use pytest to run tests locally, reporting to learner what is expected and what is actually recived in beginner friendly messages, not Pythons default error messages (before we go thrue error handling in our lessons). And there is a check.py script that runs the exercise test and produces a nice output. Tests are marked as skip all by the first one, so student removes this skip line as he progresses thrue the exercise, making it like a game - a level up game. While also providing beginner helpful messages like:

Output was expected to be:
"Hello World!"
But your output was:
"(apsolutly nothing)"

Did you use `print` to output the message?

As for actuall projects repository templates, i dont think it will clutter the project domain, but i am still thinking about that whole approach, so its not set in stone.
CLI i was planning to make in Rust so it does not need python installed on the system - you just run it and it does its job.

that way github action machine will be saved, also your account will saved from cluttering but user progress can't be saved as we don't have backend or db server for ledger..

But the user progress can be saved in multiple locations. Currently i use browsers local storage to track student progress and optional GH PAT auth to sync it to secret (users) Gist.

So persistance is not a big issue, i use Github as a backend :) if i need persistence or there are many ways to save data locally to browser

[razorblade23]

As for actuall projects repository templates, i dont think it will clutter the project domain, but i am still thinking about that whole approach, so its not set in stone.

Now that i think about it, it will be absolutely insane to maintain 20+ repositories inside the org. Basically, i have three options:

• A monorepo where i will keep the projects - downside is that students must choose a directory and i think its bad UX for students.
• Dynamic generation of template - I keep separate repository for projects (just one), but the students download their project starter from a button on a platform (engine reads the repository contents, assembles the files and puts them in .zip for student to download. This is a much better and cleaner UX for both maintainers and students.
• A separate org just for the projects templates - this is again 20+ repos to manage, just on the other "domain"... so bad UX for contributors but good for students

[JeelDobariya38]

• Dynamic generation of template - I keep separate repository for projects (just one), but the students download their project starter from a button on a platform (engine reads the repository contents, assembles the files and puts them in .zip for student to download. This is a much better and cleaner UX for both maintainers and students.

I might be missing on something, as this is very big text messageeeeee thingggg. also i am not familiar with how everything is setup (lacking domain knowledge). so, I might not be right here, but we can have cli for everything... we can have monorepo for high level concerns like one for project template-s, one for excercise-s, and cli's codebase will be smart enough to find what user needs currently and will get that for user at that specific time.. like get resources on run-time dynamically.... we can certain rules or scheme through which cli will infer where to find what when user need that..

---

this is like dream or ideal case, as what i think of cli,

User Flow

• the user flow (as in students) will be able to go to main website download the ledger and let say they need vscode up in running.
• then user can run a command call pyledgercli init which will be downloading the python, check wether user have vscode & git if not downloaded it.. and stuff this will be a itempotom command.
• then user can login, using a pyledger login can generate the report html file pyledger progress/export thier will also pyledger test 01-helloworld that what test will be run on practical excercise.. then pyledger can have a command to get template of project. and so on.. will also have command for getting starter template for project.. that like developer or mentor or what ever you calk that person in you domain model. mean that way user as in students developer as in mentor can also use cli... so "it like cli for everything"

sorry if it so vague in word it also vague idea in my head..... but this wahy we can good ux for everyone because everything is just command. (aka automated)..

and we have one place for everything... then as time pass like 10 year from now we split project into project-beginner project-backend repository that way we not premature optimising and everything remain in one place.. with 20 small repostory... we have 20 folder which more easier to maintain..

what say? it completely vague idea so please ask clarifying questions or we join in discord that what we chat instead of email / ghmessaging each other

[JeelDobariya38]

A separate org just for the projects templates - this is again 20+ repos to manage, just on the other "domain"... so bad UX for contributors but good for students

I think this will also solve if we have cli for everything contributor can have

pyledger push excercise --name ... --from-dir . --github-token ...

this make easier to add new execrise project templates for even mentors

[JeelDobariya38]

Tests are marked as skip all by the first one, so student removes this skip line as he progresses thrue the exercise, making it like a game - a level up game. While also providing beginner helpful messages like:

I don't know but utill i know we have sub-git repository. and then it will be much easier user can just git checkout level1 whete level1 is tag.. this more easier...

or wait.. we also have sub branch but that initiative & standard approch.. it violating something according to me. because if say git repository we assume it has main branch and that only it...
like if we say git repository as book. the sub branches are just in draft of intial pages that might or might not end up in final book. so user only know about final book... so we should distribute excercise into branchs... but folders like chapter one two in book... and level will be page no of specific topic in each chapter...

but if we using git tag system with execrise distribute into folder in single repository on single branch then only area of concern is that all exercise or project need to stay fully immutable once committed.. like database migration.. because if edit them once branch merger in main that diaster... so git tag system is good to have in user view but need perfect consistency and rules to ensure it doesn't crumble.. in book analogy.. it like if publish the book and then change spelling in one work in one topic and publish again then thier two book in market a lag... in our world this lag is represented by no of commit between two tags..

this also effect is seen if see things in git timeline... like if i make book publishing then some refer it in thier book i upadte my book then they upadte thier book so now thier 4 version of same book and it just matter of coincident that which you look at first you belive is fact for your life or in our world utill you hit git checkout another tag....

[razorblade23]

• Dynamic generation of template - I keep separate repository for projects (just one), but the students download their project starter from a button on a platform (engine reads the repository contents, assembles the files and puts them in .zip for student to download. This is a much better and cleaner UX for both maintainers and students.

I might be missing on something, as this is very big text messageeeeee thingggg. also i am not familiar with how everything is setup (lacking domain knowledge). so, I might not be right here, but we can have cli for everything... we can have monorepo for high level concerns like one for project template-s, one for excercise-s, and cli's codebase will be smart enough to find what user needs currently and will get that for user at that specific time.. like get resources on run-time dynamically.... we can certain rules or scheme through which cli will infer where to find what when user need that..

this is like dream or ideal case, as what i think of cli,

User Flow

• the user flow (as in students) will be able to go to main website download the ledger and let say they need vscode up in running.
• then user can run a command call pyledgercli init which will be downloading the python, check wether user have vscode & git if not downloaded it.. and stuff this will be a itempotom command.

This is not needed as the point is to learn to set up the things yourself. The point of the platform is to make real developers, not people clicking Run button and wondering why it does not work (they are in the wrong directory or something).

• then user can login, using a pyledger login can generate the report html file pyledger progress/export thier will also pyledger test 01-helloworld that what test will be run on practical excercise.. then pyledger can have a command to get template of project. and so on.. will also have command for getting starter template for project.. that like developer or mentor or what ever you calk that person in you domain model. mean that way user as in students developer as in mentor can also use cli... so "it like cli for everything"

Excercises have their own check.py script that already runs the tests on their own machine. Testing software is extremly important and this gives a chance for them to later dissamble check.py to understand how exactly does it work and to learn about code testing in later lessons.

sorry if it so vague in word it also vague idea in my head..... but this wahy we can good ux for everyone because everything is just command. (aka automated)..

Again, point of the platform is to learn to use a computer like a real developer, so while automation is good, students will need to learn to use their machine not to automate everything for them.

and we have one place for everything... then as time pass like 10 year from now we split project into project-beginner project-backend repository that way we not premature optimising and everything remain in one place.. with 20 small repostory... we have 20 folder which more easier to maintain..

Projects and excercises can live in the same repository or separate one
It does not really matter, they are mostly markdown files.

We will continue the discussion in Discord, for sure.

[razorblade23]

Tests are marked as skip all by the first one, so student removes this skip line as he progresses thrue the exercise, making it like a game - a level up game. While also providing beginner helpful messages like:

I don't know but utill i know we have sub-git repository. and then it will be much easier user can just git checkout level1 whete level1 is tag.. this more easier...

or wait.. we also have sub branch but that initiative & standard approch.. it violating something according to me. because if say git repository we assume it has main branch and that only it...
like if we say git repository as book. the sub branches are just in draft of intial pages that might or might not end up in final book. so user only know about final book... so we should distribute excercise into branchs... but folders like chapter one two in book... and level will be page no of specific topic in each chapter...

but if we using git tag system with execrise distribute into folder in single repository on single branch then only area of concern is that all exercise or project need to stay fully immutable once committed.. like database migration.. because if edit them once branch merger in main that diaster... so git tag system is good to have in user view but need perfect consistency and rules to ensure it doesn't crumble.. in book analogy.. it like if publish the book and then change spelling in one work in one topic and publish again then thier two book in market a lag... in our world this lag is represented by no of commit between two tags..

this also effect is seen if see things in git timeline... like if i make book publishing then some refer it in thier book i upadte my book then they upadte thier book so now thier 4 version of same book and it just matter of coincident that which you look at first you belive is fact for your life or in our world utill you hit git checkout another tag....

This would be just a problem for students. A monorepo with dynamic generation is a much cleaner solution.