[css-page] <page-size> Keywords

[Crissov]

TL;DR: Change JIS-B5 and JIS-B4 to JIS B5 and JIS B4.

In November 2015 it has been discussed whether to include several additional page size keywords into CSS Page. Subsequently, Japanese JIS-B4 and JIS-B5 have been added. They had already been supported by Antenna House Formatter due to strong local demand.

Prince and Antenna House actually both support additional keywords, but with inconsistencies. I didn’t find relevant documentation for Weasyprint and Vivliostyle. I quote the tables below, but I actually wonder if CSS can learn from these and do better.

size: <length>{1,2} | auto | [ <page-size> || [ portrait | landscape ] ]

auto is currently UA-dependent, but maybe it could be specified as defaulting to – if included – PA4, the (rounded) intersection of ISO A4 and ANSI A (better known as US Letter): 210mm × 280mm.

Most paper sizes for printers are specified with default portrait orientation, i.e. smaller size first, and <length>{1,2} has width first and optional height second. Some sizes are primarily used in landscape orientation and therefore they have been implemented with large size first: id-1, id-2, id-3 and ledger. Except for the first one, which is the common size of plastic cards (at 2.125in 3.37in), they are landscape aliases of other sizes: A7, A6 and tabloid, respectively. Since ledger (not tabloid) is currently specified as a portrait format, it should be clarified whether the implementations (intentionally) are non-comformant there and if it’s a bug in the spec.

International paper sizes

The ISO sizes – series A and B defined in ISO 216 and series C in ISO 269 – have been adopted by many national standardization bodies. Some have extended definitions for special purposes, e.g. the RA series from ISO 217, German DIN 2A0 or the Swedish SIS D–G series and, of course, the Japanese B series (JIS P 0138). Most of them are of limited interest to CSS authors, at least with consumer printers.

Size	CSS	Prince	Antenna House
26mm × 37mm	—	A10	—
28mm × 40mm	—	C10	—
31mm × 44mm	—	B10	—
37mm × 52mm	—	A9	—
40mm × 57mm	—	C9	—
44mm × 62mm	—	B9	—
52mm × 74mm	—	A8	—
57mm × 81mm	—	C8	—
62mm × 88mm	—	B8	—
74mm × 105mm	—	A7, ID-2	—
81mm × 114mm	—	C7	—
88mm × 125mm	—	B7, ID-3	—
105mm × 148mm	—	A6	A6
114mm × 162mm	—	C6	ISO-C6
125mm × 176mm	—	B6	ISO-B6, B6
128mm × 182mm	—	—	JIS-B6
148mm × 210mm	A5	A5	A5
162mm × 229mm	—	C5	ISO-C5
176mm × 250mm	B5	B5	ISO-B5, B5
182mm × 257mm	JIS-B5	—	JIS-B5
210mm × 297mm	A4	A4	A4
229mm × 324mm	—	C4	ISO-C4
250mm × 353mm	B4	B4	ISO-B4, B4
257mm × 364mm	JIS-B4	—	JIS-B4
297mm × 420mm	A3	A3	A3
324mm × 458mm	—	C3	ISO-C3
353mm × 500mm	—	B3	—
420mm × 594mm	—	A2	—
458mm × 648mm	—	C2	—
500mm × 707mm	—	B2	—
594mm × 841mm	—	A1	—
648mm × 917mm	—	C1	—
707mm × 1000mm	—	B1	—
841mm × 1189mm	—	A0	—
917mm × 1297mm	—	C0	—
1000mm × 1414mm	—	B0	—

I believe it could be beneficial to have <paper-size> accept two keywords, an optional one for the system (e.g. JIS) and a mandatory one for the size (e.g. B4).

• standard | auto (default, depends on the size)
• iso | international
• local
  • jis | japan | japanese
  • ansi | us | america | american | arch

One question is whether iso legal and international letter would compute to A4 and US A4 to letter.
Another question is whether UAs with local settings for Japan should equal standard B4 to JIS B4.

Other metric paper sizes

Size	CSS	Prince	Antenna House	elsewhere
53.98mm × 85.6mm	—	ID-1	—	creditcard
100mm × 148mm	—	—	Hagaki	postcard
110mm × 220mm	—	—	ISO-DL	DL, DLE
210mm × 280mm	—	—	—	PA4, L4
210mm × 330mm	—	—	Folio	F4
215mm × 280mm	—	—	—	P4
215mm × 305mm	—	—	—	RA4

• ID-1 is actually metricated 2⅛in × 3.37in.
• DL is an envelope-only size, fitting folded A4, but Wikipedia calls it “DLE” to distinguish it from ⅓ A4 “DL” (99mm × 210mm).
• PA4 or L4 is A4 cut to letter height or letter cut to A4 width (if letter equalled P4 instead of ANSI-A).
• Folio is US-Folio cut to A4 width and Wikipedia has it under the lemma F4. ISO 217 RA4 is, probably by coincidence, roughly 8.5in × 12in, i.e. between letter and legal or folio/us-folio.
• Canadian standards P1–P6 have US sizes rounded to the closest half-centimeter; CAN P4 is rounded letter or ANSI A.
• There are some inofficial ‘plus’ sizes, e.g. ‘A4+’, mostly for photo printing.

American standard paper sizes

Size	CSS	Prince	Antenna House
5.5in × 8.5in	—	US-Statement	Statement
7.25in × 10.5in	—	US-Executive	Executive
8in × 11in	—	US-Government	Government-Letter
8.5in × 11in	letter	US-Letter, ansi-a	Letter
8.5in × 13in	—	US-Folio	—
8.5in × 14in	legal	US-Legal	Legal
9in × 12in	—	arch-a	—
11in × 17in	ledger	US-Ledger, ansi-b	Ledger
		US-Tabloid	Tabloid
12in × 18in	—	arch-b	—
17in × 22in	—	ansi-c	C
18in × 24in	—	arch-c	—
22in × 34in	—	ansi-d	D
24in × 36in	—	arch-d	—
30in × 42in	—	arch-e1	—
34in × 44in	—	ansi-e	E
36in × 48in	—	arch-e	—

arch() could be defined as rounding up both sides to the next multiple of 6in, or of 3in for measures smaller than, say, 10in.

Other American paper sizes

The sizes foolscap, crown, demy, royal, imperial come in three variants each: octavo has the longer side of quarto halved and folio has its shorter side doubled. Two of these could be implemented as functions or modifier keywords for all paper sizes, so that either quarto(A4) would equal A5 or that octavo A5 was A7. half and quarter should work equally well (with a respective note on traditional terminology) and then double or quad/quadruple would also be possible, but maybe that’s not worth the hassle at all.

I’m not sure I like the alternative idea of folio(A4, 0.5) = A3, folio(A4, 2) = A5 and either folio(A4, 3) or folio(A4, 4) = A6. (Without the comma if you want and could be called fold() instead.)

Size	CSS	Prince	Antenna House
4.25in × 6.75in	—	foolscap-octavo	—
6.75in × 8.5in	—	foolscap-quarto	—
8.5in × 13.5in	—	foolscap-folio	—
5in × 7.5in	—	crown-octavo	—
7.5in × 10in	—	crown-quarto	—
10in × 15in	—	crown-folio	—
5.625in × 8.75in	—	demy-octavo	—
8.75in × 11.25in	—	demy-quarto	—
11.25in × 17.5in	—	—	—
5.75in × 9in	—	—	—
9in × 11.5in	—	medium-quarto	—
11.5in × 18in	—	—	—
6.25in × 10in	—	royal-octavo	—
10in × 12.5in	—	royal-quarto	—
12.5in × 20in	—	royal-folio	—
7.5in × 11in	—	imperial-octavo	—
11in × 15in	—	imperial-quarto	—
15in × 22in	—	imperial-folio	—

[frivoal]

I didn’t find relevant documentation for [...] Vivliostyle

As for version 2016.7, Vivliostyle supports all the page sizes specified in the spec (and only these):
https://github.com/vivliostyle/vivliostyle.js/blob/64c0766361f87ac9c8c8ddf6aef83a60931da715/resources/validation.txt#L297 : a5 | a4 | a3 | b5 | b4 | jis-b5 | jis-b4 | letter | legal | ledger. The dimensions match those specified in https://drafts.csswg.org/css-page/#typedef-page-size-page-size

Since ledger (not tabloid) is currently specified as a portrait format, it should be clarified whether the implementations (intentionally) are non-comformant there and if it’s a bug in the spec.

I believe the spec is clear that all formats are defined in portrait orientation, and Vivliostyle's implementation conforms to that.

I believe it could be beneficial to have accept two keywords, an optional one for the system (e.g. JIS) and a mandatory one for the size (e.g. B4).

I don't think I'd be in favor of this approach. It makes it possible to specify paper sizes that don't actually exist (jis ledger?), and while we could certainly define them, I don't see the benefit.

As for including more paper sizes in the spec, it is very cheap to implemented and there may be occasional demand, so we wouldn't be opposed to it, but it is also very easy for authors to work with actual dimensions if they don't have the keywords, so the need isn't that pressing either. We have yet to feel market pressure in supporting these other sizes, but if/when we do, we'll be sure to report back to the CSSWG.

It was different for JIS-B5 and JIS-B4, since there's more demand for these than for the unprefixed (implying ISO) B5 and B4, so adding there helped alleviate some confusion.

[dauwhe]

We produce many hundreds of books per year using Prince. Essentially none of them use any of the page sizes listed in any comment above. It's so very easy to type size: 6in 9.25in that I don't see a strong need for having lots of keywords.

[Crissov]

Correct me if I’m wrong, @dauwhe, but book sizes are much less standardized than loose sizes and they tend to use round values (or vulgar fractions with inches). The ISO (and JIS) sizes, on the other hand, have awful numbers to remember, because they’re designed to keep the same, irrational aspect ratio, so keywords are very useful for them.

[dauwhe]

@Crissov agreed that keywords for ISO/JIS are important; I think they're less important for foolscap-quarto and friends :)

[Crissov]

I agree that Prince’s keywords in the last table should not be included in the specification.

I’m somewhat growing fond of the fold or folio (function) syntax I mentioned, though, because it’s cleaner than using calc(), can help avoid decimal representations of 8th or 16th fractions and it would instantly cover the complete ISO, JIS, ANSI and arch series as soon as one size keyword is supported.

Personally, I don’t care whether foolscap, crown, demy, royal and imperial (without hard-coded suffix) were included.

[fantasai]

Based on the discussion here, it seems like nothing is urgent to add other than JIS-B4 and JIS-B5, which have been added already. I agree with Florian that unless there is specific demand for certain keywords, we should not add them pre-emptively. Therefore closing this issue.

[fantasai]

For the record, Shinyu Murakami writes:

My opinion is same as Dave Cramer and Florian, we should keep the list minimum.
We need jis-b5 and jis-b4 only because (ISO) b5 and b4 are already defined and those would confuse Japanese users.

And I'm inclined to defer to his position.

[Crissov]

I can live with jis-b4 being kept instead of my suggested jis b4, i.e. my original proposal being rejected.

However, @frivoal and @frivoal actually (also) wrote: “it is very cheap to implemented and there may be occasional demand” and “keywords for ISO/JIS are important”. The dimensions of an ID-1 business card or an A6 postcard are not easier to remember than of either of the B4s. It’s actually hard to remember which standard sizes out of the ISO series do have a keyword, although they are of course the most frequent physical sheet formats used for office printing. It’s different for inch-based sizes without √2 aspect ratio.

Why should authors be bothered with looking up or remembering the dimensions of standardized sizes? CSS code also becomes more readable to future editors with keywords.

As shown in the tables, there are interoperable implementations of additional keywords a6 and b6, but keywords for ISO sizes C6, C5, C4 and C3 (as well as ANSI sizes C, D and E) differ among vendors. Standardization makes sense in both cases.

[Crissov]

This is your annual reminder that keywords for the international standard sizes with an aspect ratio of the square root of two make absolute sense, i.e. everything from A0 through C10, including JIS-B0 through JIS-B10. There seems to have been a consensus about that above.

[frivoal]

I think that there's a demand problem. While this not particularly hard to add, it still needs to be specified, tested, implemented across the board, etc. So that's still a non zero effort across for many different groups.

I expect that people printing outside of the A3 - A5 range actually know how big their paper is, usually, and that the pain may be very limited for them.

If browsers add support for these, adding them to the specification will be fairly trivial. But the hard part is convincing browsers individually that this is worth doing, and putting it in the spec does very little to that end, especially since what these should do is fully obvious.

I suggest you go open bugs on individual browsers, rally around developers wanting these to vote these up, and generally work at making the demand for this feature more visible. That will have a much better chance at making it happen than pinging this group.