-
Notifications
You must be signed in to change notification settings - Fork 0
Home
VO-CLOUD dev edited this page Nov 6, 2015
·
4 revisions
Every worker will be called with a directory where to output its files. The caller of the worker is responsible for creation of the output directory. All the files have to have standard names
- data.csv
- meta.xml (follows VOTable specification)
We split metadata from data because in case of usage of a distributed filesystem, every worker needs a copy of metadata and therefore they have to be accessible independently of data.
VOCLOUD expects all intermediate files to be textual CSV (comma separated values).
The first column must contain ID or unique reference to the source. The following columns must contain data followed by optional class (or label in other words). VOCLOUD put no constraints on the optional following values.