[CT-2174] [Regression] Package max version range throwing error

[fivetran-joemarkiewicz]

Is this a regression in a recent version of dbt-core?

• I believe this is a regression in dbt-core functionality
• I have searched the existing issues, and I could not find an existing issue for this regression

Current Behavior

On the latest 1.4.2 version of dbt-core when installing a dbt package with the range including a range with a max of 0.10.0 you will receive an erroneous failure. For example:

packages:
   - package: fivetran/google_ads_source
     version: [">=0.9.0", "<0.10.0"]

Will result in

Version error for package fivetran/google_ads_source: Could not find a satisfactory version from options: ['>=0.9.0', '<0.10.0']

However, the latest version of fivetran/google_ads is 0.9.0 and should be fine with that version range.

If I then adjust the range to just not include 0.10.0 it will succeed:

packages:
   - package: fivetran/google_ads_source
     version: [">=0.9.0", "<0.9.9"]

Result

22:28:23  Installing fivetran/google_ads_source
22:28:24    Installed from version 0.9.0
22:28:24    Up to date!

Expected/Previous Behavior

The range would succeed with 0.10.0.

Steps To Reproduce

1. Install the google_ads_source package with the following range

packages:
   - package: fivetran/google_ads_source
     version: [">=0.9.0", "<0.10.0"]

2. Run dbt deps
3. See the following error:

Version error for package fivetran/google_ads_source: Could not find a satisfactory version from options: ['>=0.9.0', '<0.10.0']

Relevant log output

Version error for package fivetran/google_ads_source: Could not find a satisfactory version from options: ['>=0.9.0', '<0.10.0']

Environment

- OS: Mac OS
- Python: 3.9.6
- dbt (working version): 1.4.1
- dbt (regression version): 1.4.2

Which database adapter are you using with dbt?

bigquery

Additional Context

Here is a slack thread where a user experienced a similar issue when trying to install the fivetran/ad_reporting package.

[jtcohen6]

@fivetran-joemarkiewicz Thanks for opening!

I'm guessing this must be related somehow to the change in #6838. Taking a look

[jtcohen6]

dbt-core/core/dbt/semver.py

Lines 154 to 158 in 5ddd408

	else: # major/minor/patch, should all be numbers
	if a > b:
	return 1
	elif a < b:
	return -1

Drop a breakpoint after line 154:

            else:  # major/minor/patch, should all be numbers
                if key == "minor":
                    first = a
                    second = b
                    import ipdb; ipdb.set_trace()

ipdb> first
'9'
ipdb> second
'10'
ipdb> first > second
True

should all be numbers

let's make sure of it :)

[fivetran-joemarkiewicz]

Brilliant, @jtcohen6 thanks for working so quickly to address this! 🎉