Document: Tool datatypes_conf.xml

[innovate-invent]

Looking at the toolshed there appears to be some mechanism in Galaxy that allows to distribute custom datatypes with a tool. I can't find any documentation as to how this all works.

Does a tools datatypes_conf.xml support the full schema of the Galaxy datatypes_conf.xml?

How are these merged?

When installing a tool not from the tool shed, how is the tools datatypes_conf.xml referenced?

When installing a tool from the tool shed, does it automatically locate the tools datatypes_conf.xml?

Are there path requirements for a tool datatypes_conf.xml within the tools repository?

[martenson]

All datatypes should be defined in Galaxy, making them installable from TS was an experiment that has proven to be short-sighted. See #8531

[innovate-invent]

How do you distribute a tool with a custom datatype without being dependent on a future galaxy version?

[martenson]

You don't. Just add the datatype to Galaxy codebase, we will merge it.

[innovate-invent]

That requires some jenky workarounds then. You literally have to wait for everyone to update their Galaxy version before they can use your tool.

[martenson]

Correct, or for local usage run a Galaxy fork.

[innovate-invent]

That will only encourage people to write tool wrappers that output 'txt' files rather than properly labeling the structured text based format that it actually outputs.
There is also the issue that many bioinformatics tools invent their own crappy formats. Are you going to merge data types that only apply to specific, obscure, tool suites?

[martenson]

You can explore how we handle new datatypes by researching what datatype PRs look like: https://github.com/galaxyproject/galaxy/pulls?q=is%3Aopen+is%3Apr+label%3Aarea%2Fdatatypes

For the future we are separating datatypes into a subpackage together with other data-related functionality. This package won't have to follow main release schedule. https://github.com/galaxyproject/galaxy/tree/dev/packages/data