CI: Vizier results are different

[bsipocz]

The remote data tests fail as the results from multiple vizier queries have changed. While some of this seems OK, others are a bit puzzling, so I would instead tag @ManonMarchand to have look.

Example traceback: https://github.com/astropy/pyvo/actions/runs/12934993430/job/36077299962

[ManonMarchand]

Thanks a lot for tagging me, it wasn't in my radar!

After investigation:

• the bibcode issue is normal, here is a linked PR : fix: remove alternate identifier #643
• the conesearch one was an upstream bug. It is fixed, and should disappear in the next days once the deployment is done.

[ManonMarchand]

I'm surprised that this still fails. This is fixed in the cds registry (see for example the entry that is tested in the docs https://cds.unistra.fr/registry/?verb=GetRecord&metadataPrefix=ivo_vor&identifier=ivo://cds.vizier/J/ApJ/727/14) but I don't know when this will trickle down to the gavo registry of registries, which harvests these pages.

[msdemlei]

On Wed, Jan 29, 2025 at 07:53:34AM -0800, Manon Marchand wrote: I'm surprised that this still fails. This is fixed in the cds registry (see for example the entry that is tested in the docs https://cds.unistra.fr/registry/?verb=GetRecord&metadataPrefix=ivo_vor&identifier=ivo://cds.vizier/J/ApJ/727/14) but I don't know when this will trickle down to the gavo registry, which harvests these pages.

For the benefit of others who may run into similar riddles: Full VO registries generally harvest incrementally, i.e., they go to the publishing registries and request only records that have changed since their last visit. They typically do full re-harvests every year or so because incremental harvesting isn't the most robust thing, but if you want your changes to propagate earlier than that, you need to tell your OAI-PMH machinery that the records have changed. I can't tell you how to do that on VizieR; if your publishing registry is run by DaCHS, it would be a call to dachs pub. For ivo://CDS.VizieR/J/ApJ/727/14, this date tickling apparently has not happened. At least I'm seeing an OAI datestamp of 2022-10-24T00:00:00Z, and that's what reg.g-vo.org has; hence, it won't propagate. Could you make sure that the datestamp becomes something like now or at least pretty close? Incidentally, you don't have to wait for any trickling; you can ask the Heidelberg RegTAP service (which is what mostly serves reg.g-vo.org) for an immediate (incremental) re-harvest using the harvest trigger service: http://dc.g-vo.org/rr/q/harvtrig/form Just paste in your Registry's ivoid (I won't promise that I'm folding case, so just use ivo://cds.vizier/registry).

[ManonMarchand]

Thank you for the explanation!