Closed josephcoombe closed 4 years ago
Maybe it should crash if analysisErrorOccurred
is False and the file doesn't exist? Thoughts?
From discussion at stand-up:
Regarding whether a PET should crash if the Objective FileRef
does not exist and analysisErrorOccurred
is False
(alternatively raise an AnalysisError
if the Objective FileRef
does not exist)
... I think analyses with Design Spaces are a reason to not do either of the above.
In the image below, the HasEngine Test Bench creates an output FileRef only if the Design Space has an internal combustion engine (an alternative / optional).
In this case, analysisErrorOccurred
can still be False.
And, I think it is unnecessarily burdensome on the user to require him to create a placeholder file for configurations w/o an engine.
The Visualizer also gets cluttered because we either have to tag "placeholder files" to be ignored
or we're displaying blank engine.txt
files in the Point Details
tab for configurations w/o an engine.
@ksmyth suggested a warning/notice when the CsvRecorder
does not find a FileRef
Currently, the Visualizer code dealing with artifacts doesn't use any prior knowledge (e.g. from mdao_config.json) about what artifacts to expect.
It just unzips every artifact folder (if present) and if it finds a file matching certain patterns - e.g. images.zip containing .png/.jpeg - then it display said file. https://github.com/metamorph-inc/openmeta-visualizer/blob/2944900921197e51b7408ff37bc85f798c4e8524/Dig/tabs/Explore.R#L686-L706
This PR worked with images.zip files. I will look at whatever other output/file-type-specific Visualizer code there may be lurking... Also test what happens if an artifact is present for some iterations but not for others.