Parsing errors

[lionel-nj]

Feature request

Is your feature request related to a problem? Please describe. At present, the validation of DELFI-germany's dataset (available here) produces parsing errors and several validators might not be invoked.

Should we we able to run validation even if some stops of shapes have bad lat/lon (for example - see this discussion re: Transit.land - https://github.com/transitland/transitland/issues/27#issuecomment-258357197) ? In other words, there might be some parsing errors that we would want to continue execution on.

(from @barbeau)

How we reproduce? Run v2.0 with the following command line input: java -jar -Xmx6g gtfs-validator-v2.0.0_cli.jar -i /Users/lionel/Desktop/datasets/delfi.zip -f de-delfi -o output

• -Xmx6g is used here to bump the maximum heap space memory and avoid OutOfMemoryError.

Observed behaviour Provide a clear and concise description of what actually happened.

Screenshots:

delfi-output.zip

Environment versions

• validator version: v2.0
• Java version: openjdk 11.0.5 2019-10-15
• OS versions: Mac OSX Big Sur

[lionel-nj]

What is your point of view on this @aababilov @barbeau ?

[aababilov]

I think that we already give enough details. We ask the user to look into report.json. I am unsure if we can maintain a more specific way to briefly describe some errors (which errors to pick? - that is the question).

I also do want users to read the produced reports - that is why we generate them.

[lionel-nj]

which errors to pick? - that is the question).

~I would say the one that caused the parsing error.~ Probably list the errors that caused a parsing error.

I think that we already give enough details.

~I agree.~ Actually, from the perspective of someone external to the project it would be easier to determine the source of the parsing error if we included information about this latest in the log output.

Should we we able to run validation even if some stops of shapes have bad lat/lon (for example - see this discussion re: Transit.land - transitland/transitland#27 (comment)) ? In other words, there might be some parsing errors that we would want to continue execution on.

Do you think that having a flag to set severity level of InvalidUrlNotice, InvalidEmailNotice, InvalidPhoneNumberNotice is advisable to this extent? (linked to https://github.com/MobilityData/gtfs-validator/pull/863/files)