From railML 2 Wiki
Jump to: navigation, search

railML® Identities

Very many railML® elements require an id attribute of the W3C-Type xs:ID (external link).

This has to be provided in order to enable references, that are very much used within the railML® schemas.

On the other hand, software tools may refer to certain railML® elements, that are not referred to within the railML® file.


The general railML® description states the following semantics for this attribute.

  • id: XML-file-wide unique, machine-interpretable identity, required for later referencing that element internally. For a detailed explanation see Dev:identities.
    XML-Datei-weit eindeutige, maschineninterpretierbare Identität, die für die spätere interne Referenzierung dieses Elements erforderlich ist. Für eine detaillierte Erklärung siehe Dev:identities.


The general railML® description states the following contraints.

  • id: xs:ID (external link), required
    a string, starting with a letter (a..zA..Z) or an underscore (_),
    followed by a non-colonized and non-spaced string consisting of letters, digits, points (.), dashes (-) or underscores (_)

Export of railML® files

Software tools, which create railML® files, are free to define id values according to the above mentioned constraints.


railML® recommends using generic values, e.g.

<formation id="d2e717" name="TGV-ICE3">...

or serial numbers with letter prefixes, e.g.

<formation id="rsf0002" name="TGV-ICE3">...

or UUID's (see UUID on Wiki banner.png (external link)), e.g.

<formation id="_f36014b1-055d-4171-91a4-2a32a3210f28" name="TGV-ICE3">...

Please note: a Nil UUID (external link) shall not be used; UUID's Version 4 (external link) is commonly used. Due to the limitations of the ID type by in railML® 2 the ID shall begin with a letter. Since this is not the case for all UUIDs, some tools are adding the prefix "_" before the UUID to fufil the constraints.

Bad style

Sometimes exporting software tools create id values, that contain semantic data, like train numbers or vehicle family codes.

<formation id="fTGV-ICE3" name="TGV-ICE3">...

Please use some more generic coding style in order to not invite the importing software to parse the id values!

Wrong style

Rufzeichen.png If the semantic information is only included in the id, the importing software would have to parse the id value in order to use it.
<formation id="fTGV-ICE3">...

This coding style hinders interoperability!

Please use a generic coding style for the id value. Look for appropriate railML® attributes in the element and put the semantics there. Otherwise use the xs:anyAttribute mechanism.

The use of this Wrong style prevents certification of the export interface.

Import of railML® files


Software tools, which consume railML® files, should use the id values only for cross-referencing between elements in the imported file. When UUIDs are used, the ids can also provide valid references across files and data sources.

In case of valid railML® files, no problems will occur.

Good style

In case of consuming "good or bad styled" id values the importing software provides routines for getting the semantics from better fitting attributes.

In case of consuming "wrong styled" id values the importing software is not required to extract semantics encoded in the id value.

Wrong style

Rufzeichen.png The importing software expects to parse the id value, e.g. with a special regular expression, in order to get the coded semantics, and fails when such information is not present in the id.

The use of this Wrong style prevents certification of the import interface.

Differentiation from other indications

Many elements that have the id attribute will also support the attributes code, name and description. Each of these attributes has a separate purpose.

  • code: Like id it is a machine readable identifier, but while an id only identifies an element within a single railML® file, so that it can be referenced from other elements in the same file, a code is designed for interoperability between data exchange partners. An id is not a stable reference: the same physical object will likely have different ids in different files (unless you are using UUIDs). A code is a stable identifier for a given element across files. While an id is syntactically required to be unique within a file, there is no such syntactical constraint on a code. However, as it is an identifier, a code should help the data exchange partners correctly and unambiguously identify an object. As a consequence, use case specific uniqueness requirements may be agreed between exchange partners. As an example, a code may be valid and unique within a national network.
  • name: A short, human readable identifier. Uniqueness is not required. Its value should help a human reader separate the object from other (similar) objects. For instance, it can be displayed as a label to the user of a graphical software tool, a header in printed information or as a column in tabular representations.
  • description: Human readable text that gives additional details not given by the name. Uniqueness is not required. It may contain remarks, explanations and hints about the contents of the object.
Birne.png Please be aware that human-readable attributes are only intended to forward displayable names and simple remarks/explanations from one end-user to another end-user. They are not to be used to contain computer-processable data. Software can display the contents of these attributes to the user without modification, e.g. as a label on an object, but must never parse or interpret the contents in any way. Note that this means that, in case of discrepancies between machine processable and human readable contents, any machine processable contents have priority over possible human interpretable contents. However, these attributes may be used to transmit complementary details to machine processable parameters.

Bitte, berücksichtigen Sie, dass menschenlesbare Attribute nur für einfache Kommentare/Erklärungen von Endbenutzer zu Endbenutzer bestimmt sind. Sie sind nicht für maschineninterpretierbare Daten gedacht und werden nicht geparst. Im Falle semantischer Überschneidungen oder Widersprüche zu maschinenlesbaren Inhalten wird diesen der Vorzug gegenüber durch Menschen lesbaren gegeben. Allerdings können durch Menschen lesbare Felder für Ergänzungen zu maschinenlesbaren verwendet werden.