Bug writing .ill files to disk

Hello there,

I’d like to report an error in the .ill files written to the disk. This has been happening to me so often that I can’t imagine that I am the first reporting this, but I couldn’t find a proper thread on this.

The following picture shows clearly what is going on: 4 similar grids of 144 points are fed into an annual illuminance recipe, which returns results for 4 x 144 points. However, if I decide to re-read the .illl files, I get 3 x 144 points and one grid of 145 points. Where did this extra point come from?

This must be an error writing the results to the disk, since the results right out of the DA component are reporting the right amount of points. If I look at the files created:

One can see that one of those files is bigger than the others, hence the extra point.

This is not an isolated incident, I’ve been having this error for a while now, so I believe it can be reproduced in a different machine. I’m attaching the gh script so you can test it:
example_result.gh (278.4 KB)

My excuses if this has been discussed in another thread.

Thanks in advance!

Hi @pmcmm. It seems there is a bug when splitting the grids and writing the grid redistribution JSON file. If you set the workers to 1 (use Recipe Settings) you will see that there is 144 values because the splitting step is skipped.

@chris, I ran it locally and found in the debug folder that at this command,
honeybee-radiance grid split-folder ./input_folder ./output_folder 7 --grid-divisor 3 --min-sensor-count 200
it gives the following _redist_info.json. I assume the part in red is not supposed to be there. It adds the first value from DA_GRID_E to the end of DA_GRID_S.

If I comment out this part I get a JSON without the info in the red. You can probably better tell if this will solve it for all cases and not only this case.

1 Like

Excellent work mikkel! This indeed fixed the issue, the phantom point is gone.

I will keep working on this, and will report if I find any issue.

Thanks for getting to the bottom of this @mikkel . Are you able to push a fix to honeybee-radiance-folder for this case?

I trust your judgement about whether we should just remove those lines or add some other check before running them. @mostapha originally put together that method that evenly splits the grids and writes the dist_info JSON. So he’s probably the best to review your PR or to say if this is the correct type of fix.

1 Like