-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Rc v1.3.0 #4
base: main
Are you sure you want to change the base?
Conversation
When compiling pipelines into something runnable, instead of getting producers ( remote and literal ) into a function that's immediately callable, use a function that returns a channel that will spawn some producers before closing. With this channel we can feed literal producers as normal, while allowing us to get remote producer descriptors during runtime, without any timeout. Refactor RunPipeline to collect producers from the spawner channel in chunks, running them in the order received, until the spawner closes ( which we test for by looking at whether or not the next producer is nil ) Refactor the pipeline testcase struct to specify how many producers to run in parallel so that we can test for that, testing for non-wholly divisible and overflowing parallel counts
This reverts commit 471d1d6.
Make the struct field name consistent across pipeline descriptors and compiled pipelines Fix an error where I just forgot to prop it up with other values LOL
Codecov ReportAttention: Patch coverage is
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## main #4 +/- ##
==========================================
- Coverage 48.25% 47.02% -1.24%
==========================================
Files 22 18 -4
Lines 1005 942 -63
==========================================
- Hits 485 443 -42
+ Misses 464 458 -6
+ Partials 56 41 -15 ☔ View full report in Codecov by Sentry. |
Split parsing code into more specific files Get rid of unused code Get rid of defaultCtx, just use an empty one insteads
Pass library.Ctx to pipeline parsing so that we can use the functions that return mover descriptions SpecMap replaced with a []*Spec, specs are now only named by themselves
Since we get the mover options as cty values that are already decoded, the hcl body decoding was pretty redundant. In the future, we'll no longer make a parser, we'll just decode with regular `cty:` tags Building pipelines no longer need an evalctx. The one responsible for meta-producers needs the library context
Drop support for `psy:` struct tags, no longer use the fork providing FromCtyValueTagged
Remove pipeline blocks from the spec, replacing it with implicit decleration - that is, all of the movers of a particular group of files are implied to be of the same, single pipeline.
Since the functions kinda morphed into the same thing
Pipelines now live in `group {}` blocks. IDK if this is better than the `pipeline {}` blocks we started with, but I think so? When we want to build an executable pipeline, we monify the group. Still need to make the groups filterable with the cli
Move some more specific code for transforming pipeline and file groups into main Split out code that parses values + pipelines into full, final name TBD Move MonifyGroup into main
Create a file type in parse that has all the parsing stuff for files attached to it. Move the file group parsing stuff to its own type that implements file
The "root group" is inserted at the front of the groups list
Allow `GroupDesc` to be filtered by group names
print error in hcl diag
Use all of the producers described with produce-from blocks rather than just the first Use the standard cfg parsing entrypoint for parsing produce-from results and get rid of the partial function
closes #1 #3
Rebases
single-pipeline
to includefaster-meta-producers