Re: [lmi] GUI unit test failures after the recent changes

[Greg Chicares]

On 5/13/22 16:02, Vadim Zeitlin wrote:
> 
>  Just in case it has somehow gone unnoticed, the recent changes (I'm not
> sure which one exactly as I didn't have time to debug them yet, but it must
> be one of the 4 of the latest commits as of time of this writing, i.e.
> b90777476..fc01b15e6) have broken the GUI unit tests that now fail with:
> 
>       calculation_summary: started
>       Assertion 'materially_equal(bourn_cast<double>(irate), rate * radix)' 
> failed.
>       [ul_utilities.cpp : 152]
> 
> (see e.g. https://github.com/let-me-illustrate/lmi/runs/6420838991 for the
> 32-bit build or https://github.com/let-me-illustrate/lmi/runs/6420839080
> for the 64-bit one).

I hadn't noticed--it works here:

/opt/lmi/src/lmi[0]$git --no-pager log -1 --oneline 
3dc4f293f (HEAD -> master) Note a refactoring opportunity

/opt/lmi/src/lmi[0]$time ./nychthemeral_test.sh                   
LMI_TRIPLET = "x86_64-pc-linux-gnu"
LMI_TRIPLET = "x86_64-w64-mingw32"
LMI_TRIPLET = "i686-w64-mingw32"
./nychthemeral_test.sh  2438.16s user 248.37s system 293% cpu 15:15.76 total

>  Independently of the actual problem itself, there is also a problem with
> assert reporting, as they currently show a message box which means they
> remain stuck forever (or until 6 hours of the very generous time quota of
> GitHub Actions expire). This is something I will try to fix a.s.a.p. as
> these tests are not supposed to require any human interaction

Good.

> but I don't
> think I should try fixing the real problem because you probably already
> have something in mind of it

I haven't seen it and cannot reproduce it.

But I had already followed the directions in the email I sent
just a moment ago; unless by clairvoyance, you mightn't have.
Sorry. I hope that fixes it. Thanks for finding this: it shows
that I need to fix any script that uses the old version.