bump minor version

[bjarthur]

@kimikage had described the master branch as "dev". tests pass though. so we'll release it into the wild. let us know if any downstream dependencies break, and use v0.8.6 in the meantime.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.42%. Comparing base (cadbbc1) to head (4405708).
⚠️ Report is 15 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #325      +/-   ##
==========================================
+ Coverage   92.92%   96.42%   +3.50%     
==========================================
  Files           7        7              
  Lines         735      784      +49     
==========================================
+ Hits          683      756      +73     
+ Misses         52       28      -24     

☔ 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.

[bjarthur]

from google antigravity:

Here are the key breaking or behavioral changes you should be aware of when upgrading from
v0.8.5 to v0.9 :
──────

1. Silent Overflow in Normed Multiplication (Breaking Change)

• In v0.8.5: The * operator on Normed types was explicitly overridden to use checked
arithmetic ( checked_mul ). If a multiplication overflowed, it immediately threw an
OverflowError .
• In v0.9: This override was removed. Multiplication on all fixed-point types (including
Normed types) now defaults to wrapping arithmetic ( wrapping_mul ).
• Impact: Overflows during multiplication will now silently wrap around instead of throwing
errors. Any code relying on OverflowError safety checks during multiplication will fail
silently.
──────

2. Printing & String Representations (Breaking Change)

• In v0.8.5: print(x) fell back to show(x) and output the number with its type suffix
(e.g., print(q0f7) output "-0.672Q0f7" ).
• In v0.9: print and show were decoupled. print(x) now prints only the clean numeric
value (e.g., "-0.672" ), while show handles the type suffix and type representation.
• Impact: Any user code that performs string parsing on printed representations, relies on
string(x) , or uses string interpolation ( "$x" ) and expects the type suffix (like Q0f7
or N0f8 ) to be present will break.
──────

3. Range Length Behavior (Behavioral Change)

• In v0.8.5: Calling length on step ranges of fixed-point numbers that overflowed would
return incorrect dummy values (e.g. Base.unsafe_length(...) == -1 or 0 ).
• In v0.9: These overflowing step ranges now explicitly throw an OverflowError .
• Impact: While this is a safety correction, it changes the execution flow for any code
that previously processed or absorbed those invalid range length values.
──────

4. Deprecation Warnings (Potential Future Breaks)

If users use internal or fallback APIs, they will now receive deprecation warnings in
stderr:

• typechar : Deprecated in favor of type_prefix (which returns symbols like :Q instead
of character 'Q' ). Calling typechar now warns the user.
• floattype Fallback: Fallbacks for floattype on custom real types now emit a warning
alerting the user that a future release will throw a MethodError if a type
<:AbstractFloat cannot be returned.