HDT format identifier uses only HDT version.

[RubenVerborgh]

Issue: HDT files produced by the current master cannot be read by other HDT libraries, nor an earlier version of this library, and unnecessarily so.

Starting with 1e23041, HDT files were being written with a format IRI of http://purl.org/HDT/hdt#HDTv1-1-2 . This unnecessarily breaks compatibility with existing HDT libraries, which expect the version URL to be http://purl.org/HDT/hdt#HDTv1. Since the version number indicates HDT_version.index_version.release, only the first component matters to the HDT file itself. This commit reverts the version IRI to http://purl.org/HDT/hdt#HDTv1, restoring backward compatibility with other HDT libraries. It also reverts the version detection code to simple equality.

Partially reverts 1e23041.
Related to #36.
Fixes bug reported in LinkedDataFragments/Client.js#26 (comment).

[mielvds]

Ok, looks good to me. But shouldn't we have URIs reflecting index and version as well reflected in the header?

[RubenVerborgh]

But the index is independent of the HDT file, as we've discussed in another thread. So why should it be in the HDT?

[mielvds]

So there is no link between the HDT and the index at all? Everything is done by the software?

[RubenVerborgh]

I don't think any such link should exist, given the versioning scheme, so no as far as i'm concerned. What do you think?

[RubenVerborgh]

@mielvds Is this one okay to merge? If so, I would merge and release after (well, after we've determined the version number).

[mielvds]

Well agreed, but I want to make sure first: if we did the FM index thing, did that have to be decided at HDT generation time?

[RubenVerborgh]

Yes, but for clarity, these are two different indexes. There is the .hdt.index file, which I argue should not be in the .hdt file because they are not coupled, and there is the index of literals (a dictionary), which is inside of the .hdt file and is thus coupled to it.

[mielvds]

done.

[webdata]

Hi, I also think we don't have to include the index version in the HDT dataset. Another discussion is if we should allow the OPTION to have the index within a single HDT file (suggested by Wouter, #7 (comment)). I think we could allow this option with backward compatibility by just appending the index at the end of the file and marking this in the preamble. But this is something to discuss probably after this upcoming release