Add sizes of software artifact#1259
Conversation
Size of a software artifact can be measures in different ways. Apart from conventional bytes, audiovisual content is more meaningfully measured in duration (time) or textual AI training data in number of tokens. These addition is to align with regulatory requirements, for example, EU AI Act. Property names are closely aligned with other recognized standards/vocabs. Signed-off-by: Arthit Suriyawongkul <arthit@gmail.com>
bact
added
Profile:Software
|
@bennetkl @rgopikrishnan91 @cngo20 This is from our 17 Dec 2025 SPDX AI WG call, addressing EU AI Act's Explanatory Notice and Template for the Public Summary of Training Content for general-purpose AI models. |
|
How would you know what the units are? |
Apart from the rename of
See details in #1258 |
|
Oh, I'm sorry I read this completely backwards. My apologies. This LGTM |
Signed-off-by: Arthit Suriyawongkul <arthit@gmail.com>
stevenc-stb
left a comment
stevenc-stb
left a comment
There was a problem hiding this comment.
There is some text in Description of itemCount that is unclear.
| Constituent items can be stored within a database, embedded in a container | ||
| format, or represented as encoded binaries within a single file. | ||
|
|
||
| The unit of count is not encoded within this property. |
There was a problem hiding this comment.
itemCount is inherently a dimensionless count. Your description already states:
property records the total number of discrete constituent items contained within a software artifact.
That already defines what is being counted.
Are you meaning "discrete constituent item type" as unit of count? Can we add a example that show this. like ex. "itemCount type: files" or " itemCount type: records ?
There was a problem hiding this comment.
Thanks. I put an example for "files" here #1258 (comment)
When the property is used with a `/Software/File` element where the `fileKind`
is "directory", and the unit of count is not otherwise specified in the
`description` property, the unit of item shall be "file" and the property shall
record the number of immediate child files (including regular files,
symbolic links, and subdirectories) contained within that directory.3 participants
Size of a software artifact can be measures in different ways.
Apart from conventional bytes, audiovisual content is more meaningfully measured in duration (time) or textual AI training data in number of tokens.
These addition is to align with regulatory requirements, for example, EU AI Act.
Property names are closely aligned with other recognized standards/vocabs.
Note that the renaming of
/Software/artifactSizeto/Software/byteSizeis not a breaking change as the property is introduced in 3.1 and 3.1 is not released yet.To resolve #1258 (see more background and rationale there)