Skip to content

Disable satisfied-skip-solve on 32-bit linux#268

Open
angusmoore wants to merge 1 commit intoJuliaPy:masterfrom
angusmoore:anm/32-linux-satisfied-skip
Open

Disable satisfied-skip-solve on 32-bit linux#268
angusmoore wants to merge 1 commit intoJuliaPy:masterfrom
angusmoore:anm/32-linux-satisfied-skip

Conversation

@angusmoore
Copy link

@angusmoore angusmoore commented Dec 9, 2025

Motivation: The default install of Conda on 32 bit linux uses pre-built binaries from repo.anaconda.com/miniconda; the 32 bit linux builds are very dated and do not support the --satisfied-skip-solve flag.

Change: Ignore / disable satisfied_skip_solve on 32 bit linux, in the same way it is ignored on windows (and add a linux-specific warning if it is set to true, as is the case in windows).

Detail:

The latest 32 bit linux build (https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86.sh) is 4.5.12, which pre-dates the introduction of --satisfied-skip-solve, which was introduced in 4.6.0 (https://docs.conda.io/projects/conda/en/latest/release-notes.html\#version-4.6.0).

This is a somewhat draconian approach, as someone could have a newer build of Conda on a 32 bit linux system (installed by some other means than the default Conda installer), and this change prevents them from using --satisfied-skip-solve.

Even so, I think this change is worthwhile because (a) the current state means a default install of PyCall (which sets this flag) on 32 bit linux fails (b) preventing --satisfied-skip-solve does not change correctness, just performance (I believe, I'm not an expert on conda) (c) there are probably not a lot of people hitting this code path (on 32 bit linux, with their own, newer, Conda).

An alternative would be to explicitly check the Conda version and if it is <4.6.0 , ignore satisfied_skip_solve. However, that seemed far more complex / a lot of extra code for little gain (again, given there isn't much cost to disabling satisfied-skip-solve for a likely small share of workflows)

Reference: #245 and github.com/JuliaPy/PyCall.jl/pull/1045

@codecov-commenter
Copy link

codecov-commenter commented Dec 9, 2025

Codecov Report

❌ Patch coverage is 0% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 85.22%. Comparing base (f1f4760) to head (899db5b).

Files with missing lines Patch % Lines
src/Conda.jl 0.00% 1 Missing ⚠️
Additional details and impacted files
@@           Coverage Diff           @@
##           master     #268   +/-   ##
=======================================
  Coverage   85.22%   85.22%           
=======================================
  Files           1        1           
  Lines         203      203           
=======================================
  Hits          173      173           
  Misses         30       30           

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

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

The default install of Conda on 32 bit linux uses repo.anaconda.com/miniconda; those 32 bit builds are very dated and so do not support the --satisfied-skip-solve flag

The latest 32 bit linux build (https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86.sh) is 4.5.12, which pre-dates the introduction of --satisfied-skip-solve, which was introduced in 4.6.0 (https://docs.conda.io/projects/conda/en/latest/release-notes.html\#version-4.6.0).

This is a somewhat draconian approach, as someone could have a newer build of Conda on a 32 bit linux system, and this change prevents them from using --satisfied-skip-solve.

Even so, I think this change is worthwhile because (a) the current state means a default install of PyCall (which sets this flag) on 32 bit linux fails (b) preventing --satisfied-skip-solve does not change correctness, just performance (c) there are probably not a lot of people hitting this code path.
@angusmoore angusmoore force-pushed the anm/32-linux-satisfied-skip branch from 7e971c7 to 899db5b Compare December 9, 2025 10:48
angusmoore added a commit to angusmoore/ExcelReaders.jl that referenced this pull request Dec 10, 2025
* Remove support for pre 1.10 (this may not be necessary, but seems reasonable)
* Add 1.11 and 1.12 to CI
* Bumps macOS runner to 15 as 13 is no longer supported; requests aarch64 Julia
* Set fail-fast to false to confirm only 32-bit linux CI is (currently) failing due to JuliaPy/Conda.jl#268
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