You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I wonder if we could store plots from the memory profiler as static assets (PNGs) somewhere for every test run. Combined with #267 and a more predictable naming scheme, we then could have the tooltip for the profiles show the memory profile of that test run when you hover over the mark, just as an <img>. (This might require some altair hackiness, but I'm sure @ian-r-rose can figure it out.)
I think this would be useful both when:
You're iterating on a new test and want to see more detail about what's going on. Having the memory plots automatically saved (even to the local filesystem) might be helpful.
A memory regression occurs. Being able to compare the memory timeseries before and after the regression could be helpful to get a guess at what's going on. Does it look the same, just shifted up? Or is it spikier, etc.?
This isn't top priority, just something that might be useful in the future.
The text was updated successfully, but these errors were encountered:
I'd rather store them as pickles or yaml - this would make it, for example, very easy to overlap the data from multiple runs into a single plot.
We could also migrate MemorySampler from matplotlib to altair - probably won't need a deprecation cycle.
I wonder if we could store plots from the memory profiler as static assets (PNGs) somewhere for every test run. Combined with #267 and a more predictable naming scheme, we then could have the tooltip for the profiles show the memory profile of that test run when you hover over the mark, just as an
<img>
. (This might require some altair hackiness, but I'm sure @ian-r-rose can figure it out.)I think this would be useful both when:
This isn't top priority, just something that might be useful in the future.
The text was updated successfully, but these errors were encountered: