Skip to content

Conversation

@stemann
Copy link
Contributor

@stemann stemann commented Nov 3, 2025

E.g., for bundling for i686 on x86_64.

Also,

  • Added support for being explicit about arch (if arch is not explicitly specified in PackageBundler.toml juliaup.channel field.
  • Added platform-dependent IntType to TestPackage, where Serialization fails to deserialize if WORD_SIZE differs from bundling arch to install arch.

E.g., for bundling for i686 on x86_64.

Also,
* Added support for being explicit about arch (if arch is not explicitly specified in PackageBundler.toml juliaup.channel field.
* Added platform-dependent IntType to TestPackage, where Serialization fails to deserialize if WORD_SIZE differs from bundling arch to install arch.
@stemann stemann force-pushed the feature/juliaup_channel_arch branch from 3bffdc8 to 4b13d15 Compare November 4, 2025 15:00
@stemann stemann changed the title Add support for keeping juliaup channel arch Add support for specifying arch Nov 4, 2025
@stemann stemann marked this pull request as ready for review November 4, 2025 15:03
@stemann stemann mentioned this pull request Nov 4, 2025
…tested separately

Also,
* Added a test of bundle removal, which also ensures that no juliaup channels are not left behind.
* Ensured that the last test run is for the default arch, so the produced outputs should be usable for isolated_test CI.
@MichaelHatherly
Copy link
Collaborator

if arch is not explicitly specified in PackageBundler.toml juliaup.channel field

I'm not too sure I understand this comment. So is this a feature that can be used instead of just specifying the same kind of information in the environment's juliaup.channel field? If so, why can that just be used? I'm guessing I'm missing something here.

@stemann
Copy link
Contributor Author

stemann commented Nov 10, 2025

if arch is not explicitly specified in PackageBundler.toml juliaup.channel field

I'm not too sure I understand this comment. So is this a feature that can be used instead of just specifying the same kind of information in the environment's juliaup.channel field? If so, why can that just be used? I'm guessing I'm missing something here.

The idea here is to re-use the environment across arch’s - i.e. to just have the environment juliaup.channel specify, e.g., lts, and then call instantiate/bundle with, e.g., arch=:i686 when bundling for i686-w64-mingw32 (and have PackageBundler do the arch naming mangling to make it use the juliaup channel lts~x86 instead of just lts)

@stemann stemann force-pushed the feature/juliaup_channel_arch branch from 6542e74 to 60656dd Compare January 16, 2026 12:21
@stemann
Copy link
Contributor Author

stemann commented Jan 16, 2026

Updated wrt. #101 ; the included dev container should also make it possible to test this on x86_64-linux-gnu (bundling for i686-linux-gnu).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants