test_dump_rule_file depends on OS filename sorting

[licquia]

Description

The test TestRule.test_dump_rule_file in tests/licensedcode/test_models.py appears to rely on the native OS to sort directory listings when finding rule files to compare. On my system (Debian 12), this results in the test attempting to compare a rule file for GPL 2.0 to a rule file for the LZMA license, which obviously fails. Sorting the results of models.load_rules() for each fixes the problem.

How To Reproduce

git checkout develop
./configure --dev
venv/bin/py.test -n6

System configuration

• What OS are you running on? (Windows/MacOS/Linux)
  ** Linux, specifically Debian 12
• What version of scancode-toolkit was used to generate the scan file?
  ** The develop branch (specifically, commit 1c4d6be)
• What installation method was used to install/run scancode? (pip/source download/other)
  ** Checkout from GitHub

[pombredanne]

Thanks! I wonder if there is something special with Debian 12 and the file system encoding? @AyanSinhaMahapatra we likely want to add Debian 12 as a CI target.

[licquia]

I'm not sure; my guess was that it was simply unsorted by default, since the results from the debugger showed the files being reported in what seemed like a random order. For what it's worth, it's a brand-new install (never upgraded from a previous version of Debian), and here are the mount flags pasted from sudo mount for the root FS and /tmp:

    /dev/nvme0n1p2 on / type ext4 (rw,noatime,discard)
    tmpfs on /tmp type tmpfs (rw,noatime,inode64)

[AyanSinhaMahapatra]

@pombredanne yes, we have #2731 open to also have debian in CI