Feature request: model, … file paths relative to script location

[FloMee]

Feature request: model, … file paths relative to script location

[kahst]

I added the requested path changes and also create a cfg.CODES_FILE. However, I'll leave this issue open until I tested this on Windows with multiple threads.

[kahst]

As mentioned in #4, reverted changes for now. Do we need to run the script from outside the repo folder if we want to run it from an R script?

[FloMee]

Thank's for your quick response.

Do we need to run the script from outside the repo folder if we want to run it from an R script?

Yes, that would be great as I don't want to set my working directory for every R session to the directory of the repo.

I want to be able to distribute the executable without the remainder of the repo.

I totally agree.

Yet, what I did was add the latest checkpoint and also the codes files to the dist folder.

Copying the checkpoints and the codes files into the file of the executable in addition to the relative paths is exactly what I intended to do. With this setup I should be able to call the executable from everywhere.

Sorry for mixing up the two issues…

[kahst]

Ok, done :)

Can you please check and see if everything works as expected? Closing for now.

[FloMee]

I have to reopen the issue as there is still something missing - sorry!

Of course we need to add another line to adjust the codes file path as well.

cfg.CODES_FILE = os.path.join(os.path.dirname(os.path.abspath(sys.argv[0])), cfg.CODES_FILE)

And the loadLabels function has to be called explicitly with the cfg.LABELS_FILE variable as the default variable won't be overwritten.

cfg.LABELS = loadLabels(cfg.LABELS_FILE)

This should do the trick ;-) Thanks for your adjustments and all your effort for this project!

[kahst]

Good catch. I also added relative paths for cfg.TRANSLATED_LABELS_PATH and cfg.ERROR_LOG_FILE. Additionally, I added path transformations to other scripts, too. Closing again :)