I would add
- data can be textual or binary - binary data uses a variety of byte orders and bit orders, and numbers use base 2 and base 10 packed in various ways with various conventions for numeric sign indications. - Not all data cares about byte boundaries, and there are 5-bit, 6-bit, and 7-bit character sets. - data can have alignment constraints depending on data type ________________________________ From: Costello, Roger L. <[email protected]> Sent: Wednesday, July 3, 2019 9:55:19 AM To: [email protected] Subject: Here’s one reason why I think DFDL is great Hello DFDL community, Below is one reason why I think DFDL is great. Is there anything you would add or change? /Roger DFDL is a language for describing the form of digital data. The DFDL language represents a distillation of humankind's last 60 years' experience with formatting data. The DFDL language is a compilation of ways that humankind has deemed useful for formatting data. DFDL is agnostic to any particular file format; with DFDL nearly all file formats can be represented. For example, humankind has decided that: - data values may need to be framed (surrounded/outlined/delimited) to distinguish one value from another - framing symbols may need to be "escaped" so that the normal interpretation of a framing symbol is disabled - there are nil values and empty values ... and those concepts are different - the indicator of a nil value may be in-band or out-of-band - default values may be associated with empty values - the region for a value may be fixed. A region that is not fully filled by the data may be padded and/or filled - there may be a single occurrence of data (scalar) or a repeating occurrence of data (array) - data comes in different types -- integers, strings, booleans, date/time, etc. -- and it is beneficial to annotate data by their type
