Update README.md

[adf-ncgr]

added some details about CHECKSUM files, basic instructions and rationale

[adf-ncgr]

@StevenCannon-USDA let me know if these changes conform to your understanding of metadata management. @el239 let me know if anything seems unclear

[StevenCannon-USDA]

Looks good to me, but could be expanded. I suggest the following (more or less; please season to taste):

---

Several types of files should be checked for valid format and contents prior to pushing to the datastore-metadata repository:

  validate.sh description_genus_species YAML_FILE
  validate.sh description_genus         YAML_FILE
  validate.sh readme                    YAML_FILE

The metadata files in each collection should also be compressed and indexed, using the following script, which is found in the scripts/ directory of this repository:

  compress_and_index.sh COLLECTION

Also, calculate md5 sums for the files in the collection with the following script - again, found in the scripts/ directory:

  mdsum-folder.bash COLLECTION

For additional description of protocols related to data curation for the data store, see https://github.com/legumeinfo/datastore-specifications/tree/main/PROTOCOLS

[adf-ncgr]

Thanks @StevenCannon-USDA I'm happy to make changes but just so I'm clear you are referring to scripts in the datastore-specifications repo, correct? And is the validate.sh script you reference this one (which for some reason isn't under scripts and lacks an extension)?:
https://github.com/legumeinfo/datastore-specifications/blob/main/validate

thanks for the reminder about the detailed PROTOCOLS writeups

[StevenCannon-USDA]

... you are referring to scripts in the datastore-specifications repo, correct?

Right. My mistake.

And is the validate.sh script you reference this one (which for some reason isn't under scripts and lacks an extension)?: https://github.com/legumeinfo/datastore-specifications/blob/main/validate

No; I see that the script that we call is at /usr/local/bin/validate.sh (not under version control, afaik). It is a wrapper to the one that you mention, at the root of the datastore-specifications directory. Probably not the optimal organization ...

[adf-ncgr]

OK, thanks. I added that script to the repo and added some of the further details you suggested. Will go ahead with the merge but feel free to revise if anything seems inaccurate/insufficient (I skipped the bit about compressing/indexing since that's more specific to certain types of collections, I think- and probably already covered somewhere under the linked PROTOCOLS I'm guessing).

[el239]

This script just kept printing the usage at me, I must be doing something wrong. But, I was able to validate the yaml with executable in the parent directory as per its usage example:
./validate readme.schema.json YAML_FILE

[adf-ncgr]

thanks, were you running this on the ISU system or on one of the NCGR LIS servers? I think there's a hard-coded path in that script that may be throwing things off if not run on the former, but it should be pretty easily fixed. I put the script in the repository without thinking about it too much, but TBH I didn't actually know it existed until Steven mentioned it...

[el239]

I was on one of our own, I guess that must be why.

[adf-ncgr]

OK, I think I fixed the hardcoding issue. I think that was indeed causing the behavior you described with it just printing usage. In any case, the alternate (rawer) validation method you figured out should be equivalent, the validate.sh script just does some bit of wrapping of that.