Why is Foundry Code Repo building using an older library?
My build works fine in preview mode but is failing when building and is highlighting an error in an old tag version of the shared library that I use in my code.
I have two version of the shared library Tag (2.2.0, 2.2.1) and I pinned 2.2.* as version for the run.
The build should normally work fine as in preview mode since version 2.2.1 has no error.
I believe the problem is related to a wrong, different use of versions of a library between build and preview. Why do I have this behavior?
By specifying 2.2.* as run version, the algorithm will try to select the most recent version of the library (using tag increment number, 2.2.1 in your case) to solve the environment. But if it appears that the most recent compatible version with all dependencies is 2.2.0, this will be used for the build.
This may differ from the preview where the algorithm can select version 2.2.1.
It therefore seems that in your case the algorithm has selected version 2.2.0 instead of version 2.2.1 (which has no error ) for the build.
To check which version your build is really used, you can go to your .lock file (which lists all the versions of the packages used for the build). See screenshot below for package fields-lib for instance.
If you want to use version 2.2.1, you can force it in your meta.yml.
And when you pin a specific version that is supposed to be used in the preview as well. If it is not the case an advice is to click the “refresh dependencies” text then code assist will relaunch and re-resolve the envrionment with conda (which is slow, but should respect the pin)
So to debug this kind of issue you can :
- Look at the Lock File to find out which version of your libraries are really used
- Pin the version you want to use as workaround
- Can I export .csv (in Gzip) from Palantir but only the first part has headers?
- How can I save filter and variable statuses in Workshop?
- How do you avoid sorting when writing partitioned data in Spark on Palantir Foundry?
- How to delete a file from dataset view in Foundry?
- Refresh platform data in tables after action
- Is it Possibility to change font type in the Workshop module?
- Optimal Data Sync Strategy for Palantir Foundry: Handling Incremental Updates Based on Timestamps
- Importing RIDs from a dataset with column RIDs with Palantir Foundry Code Repository
- Palantir Foundry - How to load PDF files from Compass folder into code repository transform
- Hide tabs in workshop module
- Rest API: Is there any endpoint that provides dataset lineage info?
- What object types are avaliable for export on <hostname>/phonograph2-export/api/?
- Using a Code Repo to call a Webhook created in Palantir Foundry
- Delete all rows with duplicate values in column in Palantir Foundry Countor
- Palantir Foundry - Making http request and capturing JSON response
- How to publish a Spark ML pyspark.ml.PipelineModel object in code repositories?
- Pass a whole dataset contains multiple files to HuggingFace function in Palantir
- Is there any way to create a generic typescript function that works in different places with two different objects?
- Reference Workshop App Input Widget Value Externally (in a backend process e.g. Python Transform)
- How to filter out specific rows of dataframe in pyspark?
- How to replace output's column values with input's values in Foundry?
- Palantir Foundry - can we add object attachments in python transform?
- How to convert a Dataset of PDFs to a Media Set?
- Palantir Foundry: Composite-Key for DataSet-Rows / Object-Instance Primary Key
- How do you merge raw data feeds and connections into one in Palantir Foundry?
- Palantir Foundry - run Spark-NLP library offline
- Trigger certain event after timeout
- How to revert/roll back to an earlier commit in Foundry Code Repo
- Palantir Foundry - Can we stop propagating markings in Pipeline Builder?
- Palantir Foundry - Data Connection to copy PDF files from Compass Folder to external source?