Make unreachable! ClientBuildError assertion give a descriptive error

[reivilibre]

...since I don't know why I'm getting this error, it only tells me I shouldn't be! :D

Signed-off-by: [email protected]

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.26%. Comparing base (8119697) to head (26b1157).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4016      +/-   ##
==========================================
+ Coverage   84.25%   84.26%   +0.01%     
==========================================
  Files         266      266              
  Lines       28344    28344              
==========================================
+ Hits        23880    23884       +4     
+ Misses       4464     4460       -4     

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

[bnjbvr]

Thanks for the PR. It's bad that you can hit this panic, I suppose it was put here as an internal-API error, and never supposed to be triggerable by users of the SDK. I think the code has evolved sufficiently that it's a bad idea for us to use assert_valid_builder_args at all: the error could also be a store error, or a discovery error, or, or, or…

So instead, I think we should:

• get rid of assert_valid_builder_args
• have Client::new return a Result<Self, ClientBuildError>.

Are you willing to make the changes here? Otherwise one of us can probably get to it at some point.

[reivilibre]

I can have a go at that soon, will have to happen after the mad rush to get this other stuff done that led to this issue.

BTW it was a SqliteStore(MissingVersion).... 🤷