I would always be cautious to define biological concepts in terms of linear coordinates, the cutoffs are very "human understanding" oriented round numbers that won't generalize well.

When in doubt of what a term means consult the Sequence Ontology:

Enhancer: http://www.sequenceontology.org/browser/current_svn/term/SO:0000165

Promoter: http://www.sequenceontology.org/browser/current_svn/term/SO:0000167

An enhancer does not even have to be on the same chromosome (see http://www.ncbi.nlm.nih.gov/pubmed/16873069) for an example on the monoallelic choice of olfactory receptor WARNING the article had some conclusions invalidated), and can be upstream, downstream, up to 1Mbp away. A promoter is typically in the near cis region of your gene.

The fundamental idea is that RNA polymerase subunits are recruited (or blocked) by a multitude of transcription factors, which can bind to enhancers or promoters, forming a 'mediator complex'.

Hello, I give you my personal understanding of those concepts. A promoter is the nearest in cis discrete region allowing for (promoting the) expression of a given gene. Any other discrete sequence enhancing the expression of one or more genes is an enhancer (normally, but not necessarily, in cis). More general term is regulatory element, comprising any sequence having any kind of effect on the transcription of the gene. Regulatory elements can be cis-acting (e.g. locus control region,LCR) or trans-acting (or, sometimes, both).

As the other answers point out, a genomic distance approach is not useful to define enhancers. A more rigorous definition of enhancers might integrate multiple data sources. For example:

• Identify regions interacting with promoters in long-range interaction experiments (e.g., Schoenfelder et al. 2015)
• Since these interactions could be activating, repressive, or nonfunctional, select candidate regions that have enhancer-like chromatin modifications (e.g., from genome segmentation algorithms like ChromHMM)
• Merge those regions, or cross-reference to see the quality of predictions, with functional assays like those reported in the VISTA enhancer browser
• For the genes you really care about, you would need to experimentally disrupt the candidate enhancer (e.g. with CRISPR) to see if the candidate region is required for transcription. But even if you get a negative result, it's still possible there are other redundant enhancers, in which case you'd have to disrupt them all!

As for "promoter" definition, it may depend on your question: If you're looking at PolII recruitment, you might benefit from extending the promoter region downstream into the transcript, while focusing upstream would be useful for TF binding. For histone modifications it probably depends on the particular modification.