In T2377 we identified that a deposit via the deposit client with the param`slug` resulted with an generated uuid in its origin.
This behavior should be documented in the client documentation and in the code.
There are two scenarios with the deposit:
mention deprecation of the --slug option
| rDDEP Push deposit | |||
| D7146 | rDDEPb303b7da9f57 docs: Replace references to --slug from the user manual | ||
| D7145 | rDDEP7f4d1f1743c4 docs: Add links to description of <swh:create_origin>. | ||
| Status | Assigned | Task | ||
|---|---|---|---|---|
| Migrated | gitlab-migration | T2399 Update SWORD deposit documentation about the returned SWHID | ||
| Migrated | gitlab-migration | T2391 Document `--create-origin` on the swh-deposit client for origin creation |
When we set up an explicit deposit collection with a client we configure on SWH side a root URL for the deposits, named provider_url, and whenever a new deposit comes in, we record as a reference url for that deposit the concatenation of the root URL with the value passed to the swh deposit client via the --slug option.
The --slug option isn't mandatory, therefore deposits without --slug will have a uuid to replace the slug value.
The concatenated reference url will create in the SWH archive the origin element, which is the resource location on the web (where the software artifacts were found when using the pull mechanism).
With collections like HAL or IPOL, having a correct url makes sense, since the artifacts are on the web with a persistent identifier (a HAL-ID or a DOI). In some cases, like Intel, the deposit is not publicly available on the internet or the url is not persistent, creating a synthetic url and origin with a uuid is the best solution.
An example with IPOL,
an article where the software artifact is available has as reference URL https://doi.org/10.5201/ipol.2018.236, which we split in two:
The concatenation of these two elements is the value that we show as "origin url" when you browse the code in the archive.
An example with Intel, where the software artifacts are not available on the Intel website,
We have a synthetic origin with a fixed prefix and a generated suffix:
FYI, there's a (very concise) documentation here: https://docs.softwareheritage.org/devel/swh-deposit/specs/spec-loading.html#origin-artifact (it mentions external id for now, but it will change when swh-deposit is released)
The idea was to update the information about the slug on the documentation we give to clients :
Put link to this page in the user manual guide:
https://docs.softwareheritage.org/devel/swh-deposit/specs/protocol-reference.html#origin-creation-with-the-swhdeposit-create-origin-tag