Identification, Discrimination, Sameness

Identification is universally recognized as one of the most basic functions of SCM.

It is interesting to notice that it is based on discrimination and thus bound to a choice of tools, or of criteria. An SCM tool performing identification on the basis of file names will have problems versioning directories, and will typically consider soft links as configuration items.

Another remark concerns the level at which occurs the identification: roughly, and trying to avoid overloaded terms (version), it can be at the family or at the instance level.

Once selected the SCM tool, and within the realm of management provided by it, identification must be generic, i.e. apply equally to all the managed configuration items. It offers with sameness the basis for SCM level generic semantics.

Being so essential, identification itself should be manageable, i.e. in particular, late.

-- MarcGirod - 19 Feb 2003