2.2 update : purifying Unidatab


Announcement of next Unidatab’s release.

The clarity reached by asking the right question, has consequences on the technical side, for we can now see what wasn’t necessary in the previous architecture of Unidatab. In spite of the support I could get from the writings of respected scientists, it has become clear for me that properties of a given type (symbols for us) don’t need anything like a unicity value. In version 2.2 of Unidatab, I leave this field ‘d’ of the symbol table usable, but not used by internals anymore. A user can play with this marker and maybe test theoretical hypotheses in the realm of symbols, but Unidatab doesn’t need the feature anymore. Any new symbol is simply created with its unicity set to 0 by default (1 for a format) and users modify unicity without affecting any other process. Consequently, a symbol can’t be duplicated anymore (and that former possibility was a contradiction with Unidatab’s plea for data unicity). Only their types and properties identify symbols and Unidatab better matches its own concept under this new, purified, behaviour.

I look back embarrassed of reading here and there on this blog my recommandations for a dichotomy of properties, at a time it would have disturbed my coding efforts to cast doubt on that typological requirement. In the same way, a separate title for a record isn’t sound anymore since what a record appears to be finally is a sequence of symbols, a pure structure and nothing else. Thus a first symbol on one row should perfectly play in the future what do now a title and a format on two. This last modification, however justified, is left for a later release: it will imply some function of migration of a two-rows formatted db into a single-row formatted db. But then only will Unidatab be ready for a mathematically rigorous computation of structures.

Design changes in the course of programmation are basic mistakes that professionals avoid generally, but in spite of this new proof of amateurism, the context here is part of an explanation. Let’s remember that the epistemological dispositions we have acquired while trying to build a universal mechanism for the description of reality were not at our disposal in the beggining of the project, and the answers we have now badly fit the questions then asked. No surprise then the investigation feeds the code-design back. For such a universal mechanism is in its universality a kind of scientific idealization, maybe finally adequate to no single object, and as such interesting to philosophy. In any case, it was not available as a model before the attempt was made to build it. So the questions that guided this work were only questions “of the sort” we now enjoy some answers obtained empirically, and we may say that we have even experimented how lazyness or ignorance, or technical preferences can influence a conceptual inquiry. On a philosophical ground the work is paying back.

What are the philosophical consequences of reconsidering properties as basically univocal, not following any of the dichotomy proposed, nor extrinsic nor intrinsic, nor unique in their types? This has not to be taken as pledge for apriorism, but the treatment of such ontological differences is clearly transferred from the level of properties (closer to things) to the level of types (closer to descriptions). A good type-theory will leave us able to build our concepts in any philosophical flavour, and the underlying scheme according to which properties of any type are handled, has not to be impacted by type theory. Everything in this inquiry which is at the same time technological, philosophical and personal (by default) then comes in the realm of a needed theory of structures — hopefully close to the infrastructure of any theory — of which we could find an entry point in the study of a certain set of tupples. Let’s hope the lack of skeletality in Unidatab’s main object will not impede us to gather the benefits of new amazing researches as can be found for instance in Fong & Spivak’s Seven Sketches in Compositionality.

Below a schema of Unidatab’s new exec sequence:

unid-runtime_sequence-f

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s