Feat: Follow up on source calls

[Ellpeck]

VERY work in progress, I have no idea what I'm doing (YET)

[codecov]

Codecov Report

Attention: 8 lines in your changes are missing coverage. Please review.

Comparison is base (d69018e) 71.34% compared to head (b5ddd9a) 71.64%.
Report is 5 commits behind head on main.

Files	Patch %	Lines
src/r-bridge/retriever.ts	86.20%	3 Missing and 1 partial ⚠️
src/dataflow/internal/process/functions/source.ts	95.55%	1 Missing and 1 partial ⚠️
src/cli/repl/commands/parse.ts	0.00%	1 Missing ⚠️
src/core/print/parse-printer.ts	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #609      +/-   ##
==========================================
+ Coverage   71.34%   71.64%   +0.29%     
==========================================
  Files         217      218       +1     
  Lines        7005     7082      +77     
  Branches     1085     1094       +9     
==========================================
+ Hits         4998     5074      +76     
  Misses       1720     1720              
- Partials      287      288       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Ellpeck]

am I too stupid for a rebase, I click the "rebase" button and then it does this

[Ellpeck]

@EagleoutIce as you can see, all regular tests will also fail with the same dataflow being undefined exception just due to the fact that the source parsing code exists - regardless of whether it's actually executed.

[EagleoutIce]

First, I probably want to add "importsNotUsedAsValues": "error", to the tsconfig to prevent some of the possibilities (and deal with the upcoming errors) -> #616.

Besides it may very well be a problem of you now requiring the steps in the dataflow, while the steps require the dataflow. Can you check if something like https://github.com/import-js/eslint-plugin-import is sufficient to detect this? (and if so, add it in a new issue?).
Besides, can you break this cycle?

[Ellpeck]

It looks like the no-cycle rule doesn't detect any issues with dependency cycles. I tried a few other, seemingly related rules, but none of them seemed to catch it either. Did you have a specific rule in mind?

[Ellpeck]

import type did nothing to fix the cyclic dependency issue so that's just great

[Ellpeck]

I think this is mostly finished? I might be missing something huge and just totally forgetting about it, but all the major items from the issue #605 are implemented and should be working according to the tests :)

[EagleoutIce]

Please add an afterEach hook (?) that clears the source provider so that we can be sure they do not accidentally leak into other tests (keep it local to the dataflow tests, that should suffice).

[EagleoutIce]

Looks fine to me!

[EagleoutIce]

This pull request is included in v1.4.2 (see Release v1.4.2 (Dropping xmlparsedata, Benchmark Re-Runs, and Repl Fixes)).