Problems with p.check_iterations

[ablaom]

Newer versions of ProgressMeter have been intermittently crashing code in several MLJ repositories that use it. See for example this failing build and this issue. Version 1.5 is fine, but the new check_iterations apparatus is what is causing the problem.

Here's the beginning of the stack trace:

ERROR: InexactErr
or: trunc(Int64, NaN)                                                             
Stacktrace:
  [1] trunc
    @ ./float.jl:716 [inlined]
  [2] round
    @ ./float.jl:296 [inlined]
  [3] calc_check_iterations
    @ ~/.julia/packages/ProgressMeter/l7LEt/src/ProgressMeter.jl:246 [inlined]
  [4] updateProgress!(p::ProgressMeter.Progress; showvalues::Tuple{}, truncate_lines::Bool, 
valuecolor::Symbol, offset::Int64, keep::Bool, desc::Nothing, ignore_predictor::Bool)      
    @ ProgressMeter ~/.julia/packages/ProgressMeter/l7LEt/src/ProgressMeter.jl:293
  [5] updateProgress!
    @ ~/.julia/packages/ProgressMeter/l7LEt/src/ProgressMeter.jl:253 [inlined]

It could be that the error is on our side, but we would certainly appreciate some clues as to what we might be doing to create the NaN.

cc @OkonSamuel

[IanButterworth]

I believe this should fix the issue, but it would be great if you could check this PR if you have a repro #210

[IanButterworth]

I've merged and released #210 in v.1.7.1 given avoiding the NaN is a good idea, and likely to fix this. Can you confirm whether that new release fixes the issue. Thanks

[ablaom]

Thanks so much for the prompt attention! Having a hard time reproducing the original error (one of the reasons for reaching out) but see here.

[ablaom]

I am no longer seeing a problem in our builds, and the poster of the original issue alan-turing-institute/MLJ.jl#788 (comment) is unable to reproduce. So closing.

@IanButterworth Again many thanks for the extremely prompt resolution!