French SI Brochure feedback from BIPM (pages 4-20)

[metanorma]

French SI Brochure feedback from BIPM (pages 4-20)

[opoudjis]

Actually, don't worry: I'm going to apply Unicode Normalisation to the XML generation in standoc. @Intelligent2013 I will need you to confirm that it has worked. @manuel489 don't touch anything.

[Intelligent2013]

page	comment	to fix
4	Instead of "La Brochure sur le SI", the text starts with "Ce document"	adoc
4	the accents of the French words are all at wrong places: too high, too much on the left	xslt adoc rb
4	In the postal address at the bottom "Sèvres Cedex" should be on one line	adoc
6	Bullet point 6: gamma and alpha should not be in italics	adoc
9	Possible to bring item 5.4.8 to the previous page?	xslt
9	Include the index in the table of contents	?
10	In the title the accents on "à" and "édition" seem to be placed too high	xslt adoc rb
12	At the end of the first paragraph the reference to a page 5 is missing.	adoc
14	5. line of 2.2: "unité" should be written with the same font as"valeur numérique".	adoc
15	List of constants: It seems that the text is written in bold, but the numbers are not.	?
15	2nd line above table 1: the negative exponents are not at the same height as the positive ones. Applies to many other places in the text	jEuclid?
16	5. paragraph, line 4: an alpha is missing	adoc
16	6. paragraph: the isolated « a the end of line is not very nice	adoc
16	6. paragraph, the omega should be in italics (2 x)	adoc
16	the isolated « a the end of line is not very nice	adoc
17	2.3, first phrase: "unités de base" should be in italics (2 x)	adoc
17	side note: the word "recommendations" should be in italics	adoc
17	side note: replace "Article 5" with "chapitre 5"	rb
18	3. paragraph: replace "Article 2.3.6" with "section 5.3.6"	rb
18	definition of the kilogram: the last two digits '15' of the numberical value should be separated by a small space, so that the digits are grouped by three.	adoc
18	definition of the kilogram: the negative exponent is not at the same height as the positive	jEuclid?
18	last equation: 9 192 631 770 is missing in the equation	adoc
18	last paragraph: the isolated » at the beginning of the line is not nice	adoc
19	Definitionof the ampere: the symbol for the elementary charge, e, should be in italics	adoc
19	line above first equation: the symbol for the elementary charge, e, should be in italics	adoc
19	second paragraph from bottom: the symbol for π should not be in italics	adoc
19	second paragraph from bottom: the « at the end of the line is not nice	adoc
19	second paragraph from bottom: the symbol for the elementary charge, e, should be in italics	adoc
19	last line: the symbol for π should not be in italics	adoc
20	Definition of the kelvin: the negative and postitive exponents are not at the same height	jEuclid?
20	second equation: the decimals of 6,626 070 15 are wrongly grouped. They should be grouped as in this comments.	adoc
20	paragraph below second equation: the unit J should not be separated from the value.	adoc
20	last equation: replace 273.5 with 273.15	adoc
20	replace "article 5.4.1" with "section 5.4.1"	rb

[opoudjis]

rb:

Not that BIPM have said this explicitly (and BIPM must learn that they need to!!), but it appears that instead of using "Article" for clause references, BIPM uses "Chapître/Chapter" for first level clause references, and "Section" for subclauses.

[Intelligent2013]

@opoudjis @manuel489 Please note that I've reassigned these issues to adoc and rb:

page	comment	to fix
4	the accents of the French words are all at wrong places: too high, too much on the left	xslt adoc rb
10	In the title the accents on "à" and "édition" seem to be placed too high	xslt adoc rb

It's a bug in FOP: https://issues.apache.org/jira/browse/FOP-2969

In the source xml there are a characters with accent/acutes as separated characters, i.e. for example:

In normal case - Latin Small Letter E With Acute has a hex code E9:

but in the source adoc and xml it encoded as e and hex code 301 (Combining Acute Accent)
== Preface à la 9^e^ édition

These chars should be replaced to simple characters (i.e. as one char).

Resulted PDF - comparing between old and updated (copied from word to xml directly) chars:

[opoudjis]

Wow! Yes, although Unicode processing should be dealing with that, because those two character sequences are formally equivalent, this is indeed a problem, and precomposed characters are always preferred where available. @manuel489 You are already editing the documents; can you fix this, or do you need me to?

[opoudjis]

Turns out, it is any French text I have entered, in Mac Terminal: Mac Terminal is entering decomposed Unicode text. So that applies to the YAML tables and to the boilerplate. I will introduce normalisation anyway, since this can happen at any stage.

I also am changing ALL METANORMA FILES to NFC (the correct, composed Unicode normalisation).

[anermina]

page comment to fix
4 Instead of "La Brochure sur le SI", the text starts with "Ce document" adoc
4 In the postal address at the bottom "Sèvres Cedex" should be on one line adoc

These are not part of adoc file. This appears on the page right after the front page. @opoudjis?

6 Bullet point 6: gamma and alpha should not be in italics adoc
19 second paragraph from bottom: the symbol for π should not be in italics adoc
19 last line: the symbol for π should not be in italics adoc

We can fix this by applying some other font, such as sf alpha , sf gamma and sf pi. @ronaldtse would that be fine?
I didn't apply this yet since we also have a reverse case when italics are needed, but we don't have this kind of option, since something like cc Omega would produce the same result as Omega...

12 At the end of the first paragraph the reference to a page 5 is missing. adoc

I added a cross-reference.

14 5. line of 2.2: "unité" should be written with the same font as"valeur numérique". adoc
16 5. paragraph, line 4: an alpha is missing adoc

Fixed.

16 6. paragraph: the isolated « a the end of line is not very nice adoc
16 the isolated « a the end of line is not very nice adoc

I removed blank spaces between   and «/» symbols.

16 6. paragraph, the omega should be in italics (2 x) adoc

As @manuel489 wrote in #41 (comment), this is not fixable by markup, unless we insert a literal symbol. @ronaldtse?

17 2.3, first phrase: "unités de base" should be in italics (2 x) adoc
17 side note: the word "recommendations" should be in italics adoc

Fixed.

18 definition of the kilogram: the last two digits '15' of the numberical value should be separated by a small space, so that the digits are grouped by three. adoc
20 second equation: the decimals of 6,626 070 15 are wrongly grouped. They should be grouped as in this comments. adoc

As I understand, this will be solved in metanorma/metanorma-bipm#44. We removed manual digits separation from adoc.

18 last equation: 9 192 631 770 is missing in the equation adoc

Fixed.

18 last paragraph: the isolated » at the beginning of the line is not nice adoc
19 second paragraph from bottom: the « at the end of the line is not nice adoc

I removed blank spaces between   and «/» symbols.

19 Definitionof the ampere: the symbol for the elementary charge, e, should be in italics adoc
19 line above first equation: the symbol for the elementary charge, e, should be in italics adoc
19 second paragraph from bottom: the symbol for the elementary charge, e, should be in italics adoc
20 paragraph below second equation: the unit J should not be separated from the value. adoc
20 last equation: replace 273.5 with 273.15 adoc

Fixed.

[opoudjis]

4 Instead of "La Brochure sur le SI", the text starts with "Ce document" adoc

That is deliberate by me, and I strongly urge us not to fix this. The boilerplate we are including in our generated documents is not intended just for the single document of the SI Brochure, but for all documents produced by BIPM. We cannot just insert the title of the document, because in French we don't know what the gender of the document title is (La brochure, but le rapport). And frankly, I do not believe that inserting the document title in the boilerplate is worth us adding a document title gender attribute.

4 In the postal address at the bottom "Sèvres Cedex" should be on one line adoc

It's boilerplate. Fixed.

[opoudjis]

16 6. paragraph, the omega should be in italics (2 x) adoc
As @manuel489 wrote in #41 (comment), this is not fixable by markup, unless we insert a literal symbol. @ronaldtse?

@anermina @manuel489 the italicisation of Greek letters within mathematical expressions will be addressed in metanorma/metanorma-standoc#367.

(Mathematical formatting tools default to [crude] English conventions of italicising all Greek and Roman letters in mathematical tools, and only TeX allows that to be configured differently, including to French practice, which does not italicise Greek lowercase letters. We will need to address this by forced override of italicisation in MathML.)

[opoudjis]

Updated PDF: #24 (comment)

[manuelfuenmayor]

Based on the latest PDF provided by @opoudjis: si-brochure-fr(12-01).pdf

I've made comparisons between these three files: PDF reviewed by BIPM, latest PDF provided by Nick, and original PDF.

These are the issues that don't fulfill with BIPM requirements yet.

page	comment	is issue fixed?
4	Instead of "La Brochure sur le SI", the text starts with "Ce document"	no
9	Include the index in the table of contents	no
10	In the title the accents on "à" and "édition" seem to be placed too high	no
15	List of constants: It seems that the text is written in bold, but the numbers are not.	no
16	6. paragraph, the omega should be in italics (2 x)	no
19	Definition of the ampere: the symbol for the elementary charge, e, should be in italics	no
19	line above first equation: the symbol for the elementary charge, e, should be in italics	no
19	second paragraph from bottom: the symbol for the elementary charge, e, should be in italics	no

Ping @opoudjis , @Intelligent2013

[anermina]

19 Definition of the ampere: the symbol for the elementary charge, e, should be in italics no
19 line above first equation: the symbol for the elementary charge, e, should be in italics no
19 second paragraph from bottom: the symbol for the elementary charge, e, should be in italics no

I'd just like to add that the last three issues were actually source adoc issues when BIPM reported them. In other words, this was the output:

These issues have been solved and now we have the output where e is in the same font as in other parts of the text, for which BIPM didn't add any notes. I believe they are fine with this appearance of the letter e. Yellow highlighted (corrected) is the same as red underlined (initial output).

[manuelfuenmayor]

Wow! Thanks @anermina for double-checking!
Hmm, from that point of view, yes I think you're right 😃.

The reason I marked it as not-fixed issue is because I compared it with the font of the original file and they are different:

left: generated, right: original

But yeah, that is probably because we are using a different set of fonts. Which in this case, the difference between italics and not italics is a little bit more subtle (but definitely noticeable):

above: italics, bottom: not italics

Thanks!

[manuelfuenmayor]

So, discarding the issues mentioned above, the table results in:

page	comment	issue
4	Instead of "La Brochure sur le SI", the text starts with "Ce document"	not fixed
9	Include the index in the table of contents	not fixed
10	In the title the accents on "à" and "édition" seem to be placed too high	not fixed
15	List of constants: It seems that the text is written in bold, but the numbers are not.	not fixed
16	6. paragraph, the omega should be in italics (2 x)	not fixed

[anermina]

No problem @manuel489 , I just remembered seeing this in some of the previous checks :)

[opoudjis]

p. 4: #40 (comment) . I am currently declining to implement, as I want strongly boilerplated text.

p. 9: There is currently no index anyway; metanorma/metanorma-bipm#67 is the task to create it, and it is a big undertaking

p. 10. The accents are supposed to have been fixed, by me applying Unicode Normalisation to the text generation; will need to investigate what has happened.

p. 15, 16: to investigate

[opoudjis]

p. 10, clearly I missed this file. Fixed.

[opoudjis]

p. 15. Boldface applied to text does not apply to the MathML it contains; so any boldface has to be indicated explicitly on the AsciiMath: the MathML processor simply ignores the formatting of its context. (I don't think that's the case in Word, but then, Word is not an exemplar of mathematical word processing.)

Unfortunately this means you're also going to have to insert bii(), bold italic, since the override is going to wipe out any italics already there.

Moreover, nesting of font styles tends not to work.

So you get something disgusting to data enter like:

* *la fréquence de la transition hyperfine de l’état fondamental de l’atome de césium 133 non perturbé, stem:[bb(Delta) bii(nu)_bb("Cs")], est égale à stem:[bb(9192631770) " " bb(H)bb(z)],*
* *la vitesse de la lumière dans le vide, stem:[bii(c)], est égale à stem:[bb(299792458) " " bb(m)//bb(s)],*
* *la constante de Planck, stem:[bii h], est égale à stem:[bb(6.62607015 xx 10^{-34}) " " bb(J) " " bb(s)],*
* *la charge élémentaire, stem:[bii e], est égale à stem:[bb(1.602176634 xx 10^{-19}) " " bb(C)],*
* *la constante de Boltzmann, stem:[bii k], est égale à stem:[bb(1.380649 xx 10^{-23}) " " bb(J)//bb(K)],*
* *la constante d’Avogadro, stem:[bii N_"A"], est égale à stem:[bb(6.02214076 xx 10)^bb(23) " " bb(mol)^{-1})],*
* *l’efficacité lumineuse d’un rayonnement monochromatique de fréquence stem:[bb(540 xx 10^12) " " bb(H)bb(z)], stem:[bii(K_{"cd"})], est égale à stem:[bb(683) " " bb(lm)//bb(W)],*

AND, any such overrides of style are themselves going to be overridden by the styles we insert to deal with upright Greek characters. So we can't make Greek characters boldface without more postprocessing.

AND in JEuclid (PDF), boldface is being ignored for exponents.

As a result, what I'm getting with the data entry above is:

PDF:

HTML:

I have no idea why random letters H and W are not being bolded by jEuclid.

This is enough of a disaster to push back on the use of boldface for emphasising mathematical text, which is frankly ugly to begin with. I do not think we should be proceeding with this.

[opoudjis]

p. 16.

There are two different things going on here.

First: If this were a <mi> operator, as we would expect, it would be rendered in italics. It is being rendered as a <mo> instead by the AsciiMath converter.

Second: font commands will work in the general case; e.g. stem:[ii Omega] instead of stem:[Omega]. However, any such font commands should override the font commands introduced for maths localisation. For example, we expect stem:[Omega] to render <mi>&Omega;</mi>; that is localised for French typesetting as <mi mathvariant="normal">&Omega;</mi>. We expect stem:[ii Omega] to be respected, and to cancel the override of mathvariant="normal".

This arguably annuls any font nesting we might try to introduce; but we have just seen that font nesting in MathML does not work anyway.

Addressing the second, and inviting editors to insert explicit font overrides in cases like this. The first needs further investigation of the asciimath gem, but might not be tractable.

[opoudjis]

First issue: Omega is being parsed by AsciiMath as an operator, because, well, it is an operator in mathematics: https://en.wikipedia.org/wiki/Omega#The_symbol_Ω_(uppercase_letter) ; e.g.

In number theory, Ω(n) is the number of prime divisors of n (counting multiplicity)

Greek Letters that are treated by AsciiMath as operators are: https://github.com/asciidoctor/asciimath/blob/7b48295ae5c5bfc3f4679daa9ca322f69aec1931/lib/asciimath/markup.rb

Γ, Δ, Θ, Λ, Ξ, Π, Σ, Ω

The French localisation of mathematical variables is not going to catch operators <mo>. So we have a problem with trying to do fine semantics with AsciiMath.

We can override this behaviour by postprocessing, to say that if any of those operator letters is not followed by parentheses in the same MathML expression, they should be switched to <mi>. I'm wary of spending too much time fixing these, but the blanket assumption that those capital Greek letters are mathematical operators (which are always upright) is inapplicable in non-mathematical use of Greek letters, such as in physics.

This is very convoluted; let me summarise it.

LETTER	AsciiMath RENDERING	Default Rendering Style	French Rendering Style
lowercase Roman a	mi	italic	italic
uppercase Roman A	mi	italic	upright
lowercase Greek alpha	mi	italic	upright
uppercase Greek alpha	mi	italic	upright
uppercase Greek omega	mo (INCORRECT)	upright	upright

We have the unfortunate situation that:

• Capital omega is being treated as an operator, which renders it as upright
• Capital omega should being treated as a variable, which still renders it under French rules as upright
• The expected rendering in the document in this case is as italics, which violates French rules, and observes English rules instead.

The markup stem:[ii Omega] will address the rendering in either case, and the semantics in my opinion is not a priority to fix, especially if Units markup is going to be revisited anyway. But the operator treatment of Greek letters is a gotcha in Asciimath.

[manuelfuenmayor]

Ok, italic markup for Omega symbols and M operators is applied in #59

[manuelfuenmayor]

I have no idea why random letters H and W are not being bolded by jEuclid.
This is enough of a disaster to push back on the use of boldface for emphasising mathematical text, which is frankly ugly to begin with. I do not think we should be proceeding with this.

@opoudjis, just to confirm, should we proceed with boldface markup in math environment?

[manuelfuenmayor]

One remaining issue in this thread:

page	comment	issue
4	Instead of "La Brochure sur le SI", the text starts with "Ce document"	will not fix, #40 (comment)
9	Include the index in the table of contents	not fixed, to be done in metanorma/metanorma-bipm#67
10	In the title the accents on "à" and "édition" seem to be placed too high	fixed
15	List of constants: It seems that the text is written in bold, but the numbers are not.	will not fix, #40 (comment)
16	6. paragraph, the omega should be in italics (2 x)	fixed

[manuelfuenmayor]

@opoudjis , there is another issue that weren't before and I missed to report.

An additional entry in the index, Références, that is not in the original document:

And it takes a whole page for itself.

[opoudjis]

@manuel489 The references are correctly deleted from the documents within the bilingual brochure folder.

To explain:

• A reference of a document to another document must result in a bibliography
• A reference from a document to another document within a collection results in that reference being hidden—because citations are replaced with cross-references
• So those references are hidden in the documents generated as part of collection processing—the documents generated within the bilingual-brochure folder.
• OTOH those references are NOT hidden in the French brochure when it is generated, beforehand, as a self-standing document. (si-brochure-fr.pdf outside of the bilingual-brochure folder, as opposed to si-brochure-fr.pdf inside the bilingual-brochure folder.)
• And that behaviour is correct. It is the height of absurdity for a document to reference another document only by page number, when it is being treated as standalone. By definition. The surprise here is only in that BIPM have not generated these documents as standalone.

Because the references are being deleted in the collection brochure, there is nothing for me to fix.

[manuelfuenmayor]

Ok @opoudjis, thank you for your clarification. (And sorry for the bothering.)

[anermina]

Indexing is addressed in metanorma/metanorma-bipm#67. Closing this ticket.