Project

General

Profile

Task #2096

Feature #1764: Finalize dataoneTypes schema for public release

Identifier should not be able to have any whitespace

Added by Roger Dahl over 12 years ago. Updated over 12 years ago.

Status:
Closed
Priority:
High
Assignee:
Category:
Documentation
Target version:
Start date:
2011-12-07
Due date:
% Done:

100%

Milestone:
CCI-1.0.0
Product Version:
*
Story Points:
Sprint:

Description

Identifier should not be able to have any whitespace, but that is currently not
being enforced in the schema. Identifier is a NonEmptyString800, which is a
NonEmptyString with length <= 800. NonEmptyString is a string with length > 0
and which matches "[\s][\S][\s\S]". The regex allows any string that has at
least one non-whitespace character.

[\s\S] allows newlines (unlike "."). In most strings, I don't think we want to
allow newlines, and in the strings where we do want to allow whitespace,
shouldn't it be the opposite -- that it's allowed within the text, but not in
the start or end (where it may be invisible).

I think we may want to add another string type that is used only for blocks of
text that may contain newlines.

So,
Identifier (and other names) would be len < 800 and one or more non-whitespace: "\S+"
Names that allow whitespace only within text and does not allow newlines would be: \S|\S.\S
Names that allow whitespace only within text and does allow newlines: \S|\S[\s\S]
\S

History

#1 Updated by Dave Vieglais over 12 years ago

The problem description needs some clarification.

Identifiers are non-empty strings of limited length that do not have leading or trailing whitespace or non-printing characters. The restriction SHOULD be enforced by the schema, but MUST be enforced programmatically.

#2 Updated by Matthew Jones over 12 years ago

the \s and \S patterns work for the common ASCII whitespace characters, but in XML schema they don't catch all of the non-printing whitespace characters that are available in other character codesets that could be used in UTF-8 strings. I felt it was non-trivial to write a pattern that properly excluded all of the non-printing/whitespace characters in Unicode. I still think this is the case. We could write a pattern that excludes the \s characters, but we still wouldn't know if it follows the spirit of the no whitespace, no non-printing characters specification.

#3 Updated by Dave Vieglais over 12 years ago

  • Assignee changed from Matthew Jones to Chris Jones
  • Target version changed from Sprint-2011.49-Block.6 to Sprint-2011.50-Block.6

Decision: Identifiers should not have whitespace.

  • Limit as best possible through regexp in the dataonetypes schema the structure of identifiers to prevent whitespace appearing anywhere in an identifier. Such a test may be limited to ASCII because of hte implementation of regexp evaluations by xml schema processors.

  • Ensure that programmatic checks are made on identifiers to verify that no white space is present in new identifiers added to the system (reserve, create)

#4 Updated by Chris Jones over 12 years ago

  • Status changed from New to Closed

Added the NonEmptyNoWhitespaceString800 type to the schema.

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 14.8 MB)