Re: Hydra in relation to LDF/TPF

Hello Tomasz,


The next telco is already planned for next Monday. Let us know if you will be able to join and we’ll make it a point to discuss.

Perfect, we’ll discuss internally, but someone should be there at 8pm CET


As for the relation between Hydra and TPD/LDF, personally I think that too close relation is not healthy. Being tech-agnostic, Hydra should only serve as basis for more specialized tools, defining common ground for various applications. Mixing TPF/LDF into the spec itself causes confusion and could eventually make Hydra too inflexible.

I don’t think that was ever the intention. Yes, all three specs are hosted/maintained by the Hydra CG, however there is no dependency of the Hydra on LDF/TPF whatsoever (other way around obviously yes), but I get why a mixed github issue list creates that impression :)

What I find important though regarding 'common ground for various applications’, is preventing that it becomes so generic that it no longer serves any purpose (there are many examples of specs). Keeping an implementation spec (LDF, TPF, or anything else) around (in the CG or somewhere closeby) can help Hydra stay grounded.


That said, we’ll be more that happy to re-evaluate individual cases if we can reach one of two states:

1. To include proposed features into Hydra which benefit TPF/LDF as well as other APIs
2. To build extension points in Hydra which allow LDF/TPF-specific customisation, which otherwise will not leak into the generic spec

Yes, completely agree, with a preference for 2!

Cheers,

Miel


Best,
Tom

On 9 Jan 2019, at 09:36, Miel Vander Sande (UGent-imec) <Miel.VanderSande@UGent.be<mailto:Miel.VanderSande@UGent.be>> wrote:

Hi Karol, all,

Our group has put the LDF and TPF specifications under the wings of Hydra some time ago, with the motivation that, at the time, LDF/TPF were the first applications or use cases of the Hydra vocabulary.
This had the benefit that some (practical) issues in Hydra, e.g., the unclear semantics of hydra:search, got noticed and all specifications could push each other forward.

When the activity in the CG declined, probably so has the relationship between both specifications and it seems the presence of LDF/TPF has become unclear. Now, with new leadership and momentum, it might be a good time to reassess the position of LDF/TPF in the Hydra CG. LDF issues becoming obsolete or ‘out-of-scope’ caught our attention and, while we definitely understand the reasons for separating scopes (I guess at least the github issues could be split), we better discuss first before we loose any important ties, documented issues or work.

Unfortunately we missed the last telco, but I propose to put in on the agenda for next one to see how to best proceed. Would that be ok?

Best regards,

Miel Vander Sande
Postdoctoral Researcher at IDLab, Ghent University, in collaboration with imec

AA Tower | Technologiepark 19 9052 Ghent
www.idlab.technology<http://www.idlab.technology/>
@Miel_vds