PDF/A-1: the first archiving standard, and what it settled
The first edition, published in 2005, is the one that set the rule the four editions have applied ever since: everything the document needs to be read is inside the file. The fonts are in it, the colours are described in it, the pictures are included in it. Nothing to fetch elsewhere, neither over the network nor on the machine that opens it.
- Published text
ISO 19005-1- Year of the edition
- 2005
- Version it rests on
- PDF 1.4
Summary
In 2005 a PDF could name a font it did not carry, describe a colour as whatever this particular screen called red, and point at a picture held somewhere else. All of it worked, and all of it stopped working the day the font, the screen or the address was gone.
So a set of rules was written that shut every one of those doors at once. It is short, it is strict, and it is the reason a document from 2006 written this way still opens today in Adobe Acrobat Reader (Windows, Mac, Android, iPhone), in Preview on the Mac, in Chrome, Firefox and Edge, in Okular and Evince on Linux and in SumatraPDF on Windows.
It came in two levels. The plainer one promises the page: it will look the same for as long as the file exists. The fuller one promises the page and its meaning: the document says which text is a heading and which is a paragraph, so software reading it aloud says it in the right order.
Technically
The part rests on PDF 1.4, the version of the format Adobe published before the format became an ISO standard of its own. That is what dates it: everything the format gained afterwards is outside this part by construction, which is why the later editions exist.
Two levels, written as a letter after the number. Level B asks for the page to be reproducible; level A asks for that and for a structure tree, a mapping from every drawn character back to Unicode, and the language of the text to be stated.
The claim is stated in the document's own description packet, in two entries: the part and the level. A file whose description says one thing and whose contents say another is not a matter of opinion — a checker reads both.
Clause by clause
This part is among the claims the engine writes, at both its levels. What it asks for is what the later parts ask for, plus the constructs it alone closes off: a document claiming it is refused rather than written wrong if it carries one of them.
A document that has to be claimed as this part today is best written as one of the later ones and checked against the requirement: every later part holds to the same rule — everything needed to read the document is inside the file. Which edition it takes is for the archive to state: some name the family and not an edition, as the United States National Archives do in their table of preferred formats.
The levels of this part
| Level | What it adds | Asked for over the network as |
|---|---|---|
PDF/A-1A |
The page, plus the structure of its own text, its language and a mapping from every character drawn back to the letter it stands for. |
pdfa1a
|
PDF/A-1B |
The page alone: it looks the same for as long as the file exists. |
pdfa1b
|
What this part asks for
- Every font the document draws with travels inside the file.
- A colour that depends on the device is accompanied by a statement saying which device, so it can be reproduced rather than guessed at.
- The document carries a description packet naming the part and the level it claims.
- Every annotation carries the drawing of how it is to appear.
What this part rules out
What every part of the standard rules out
On top of what every part of the standard rules out, this part closes:
- See-through paint and layers that can be switched off: both arrived in the format after this edition was written.
- A file carried inside the document: the second edition allows that.
Where this part stands here today
Both levels of this part are written, and a request asks for either by its own name. A document claiming this part is written on PDF 1.4 and is refused if it carries what the part closes: see-through paint, a part of the page that can be switched off, or a file inside it.
What an archival document written here carries is the same whichever part is claimed — the fonts, the output intent, the description packet, the report and the two independent checkers — and it is set out once, on the page that presents the standard. What every archival document here carries
The requirements, family by family
The requirements group into eight families, the same eight for every part, and each of their pages says what we write for you automatically and the little that is left for you to state. The eight families, and the page each one has
The published texts behind these rules
| Published text | What it asks for |
|---|---|
ISO 19005-1 6.1.11 |
A file specification never names a carried file: this part carries nothing inside the document. |
ISO 19005-1 6.1.13 |
The document declares no set of layers that can be switched off. |
ISO 19005-1 6.2.2 |
A colour that depends on the device is accompanied by an output intent naming that device. |
ISO 19005-1 6.3.4 |
Every font the document draws with is carried inside the file. |
ISO 19005-1 6.4 |
Nothing on the page is painted see-through. |
ISO 19005-1 6.6.1 |
No action launches a program, plays a sound or runs a piece of code. |
ISO 19005-1 6.7.11 |
The part and the level are stated in the document's own description. |