A5 paper size is (slightly) incorrect

[bruceoberg]

it appears that the size of a5 paper defined in fpdf.PAGE_FORMATS is incorrect along it's narrow dimension. fpdf's A5 narrow dimension is 1.41 points longer than the ISO defintion.

converting millimeters to points is a simple multiplcation by two factors: mm to inch (1/25.4 aka 0.039370079) and inch to point (72). combining these factors gives us a mm to point conversion factor of 2.834645669.

ISO 216 defines A5 paper has having dimensions of 148mm x 210mm. when i convert these to points via the above conversion factor, i get 419.53pt x 595.28pt (rounded to hundredths).

in the source, the dictionary for fpdf.PAGE_FORMATS lists the size for a5 as (420.94, 595.28). the wide dimension matches the caclulations above. the narrow dimension, according to fpdf, is 1.41pt longer than these calculations.

fpdf's definition has been in place since the creation of this repository. it's likely that many clients depend on it, despite its incorrectness. 1.41 points is a small enough error that it may not affect anyone.

so i can't advocate that this be changed; i just wanted to document what i found when building my own table of ISO page sizes.

[Lucas-C]

Hi @bruceoberg

Thank you for reporting this, and sorry for the delay in answering you.

I personnally think that we should fix this in the next release.
I opened PR #1703 to do so.
It includes a warning that I plan to only include in the fpdf2 release 2.8.5.

What do you think about this @andersonhc, do you agree with this plan? 🙂

[Lucas-C]

@allcontributors please add @bruceoberg for bug

[allcontributors]

@Lucas-C

I've put up a pull request to add @bruceoberg! 🎉

[bruceoberg]

thanks @Lucas-C

do you need me to approve the PR? it looks perfect.

for what it's worth, i've created a python table of most official paper sizes. it includes all the ISO 216 sizes (A/B/C x 0..10), , the ISO 217 sizes, ANSI sizes and several other official sets, most with links to wikipedia sources. i could imagine this table (MORE_PAGE_FORMATS) being included in fpdf2, but i bet some extra documentation would also be necessary.

[andersonhc]

I personnally think that we should fix this in the next release. I opened PR #1703 to do so. It includes a warning that I plan to only include in the fpdf2 release 2.8.5.

I agree we should fix it. The difference is very small so it shouldn't be noticeable to any existing user.