[NTG-context] Re: redefine space to be the same as \␣ similar to knuthian approach

18 Aug 2023

      On Fri, Aug 18, 2023 at 06:20:29PM +0200, Mikael Sundqvist wrote:
...
Hi,
It is extremely difficult to follow what you write.
On Fri, Aug 18, 2023 at 11:43 AM Carlos  wrote:
...
If I have the following, with these linebreaks as in:
{\par But a system cannot be successful if it is too strongly
influenced by a single person. {\obeylines Once the initial design is
complete and fairly robust, the real test begins as people
with many different viewpoints undertake their own
experiments.}}
Is that the complete document? What do you have in mind with obeying
lines in the middle of a paragraph?
...
and opted to load another font, other than cmr that is, a \frenchspacing
approach wouldn't be further required
cmr? Not used in ConTeXt for a long time. (And what does the changing
of font have to do with this?)
...
Bear with me here, in the current state, for example, and as long as say
«…person.␣{\obeylines Once the initial is
complete…» though feasible enough, leaves any prior \␣ at the mercy of
whatever fontsize and/or set width happens to be. And this is just plain
wrong.
What?
...
Likewise, if a word sequence such as \TeX\ occurs as in {\ss The separation
of any of these four components would have hurt \TeX\ significantly. }
Likewise what?
...
The next sentence: «If I had not participated…» does not get any
\nofrenchspacing which is equally and doubly problematic. It shows lack of
consistency. And this ought not to be an ‹either› ‹or› scenario. But
rather, an and conjunctional construct. It fails both ways.
Consistency of what? Spacing? Where? Can you make a complete example?
(You can show space amount with \showmakup[space])
...
Furthermore, with the same token, if width is specified with a
\setuplayout[width=15cm]
OK, here the game changes...
...
Anything less than 12.895pt, especifically for that use case, wwould
throw anything, particularly control sequences such as \TeX\ out
of whack, and conversely, once a value of that very pt or pica or
whatever is lowered, it brings that nonfrenchspacing right back on.
And if width increments occur, then it follows that any control sequence kerning
also gets thrown off as a result.
Of course the width influences the spacing. That is how the paragraph
builder works (and really, why it often looks good).
...
It seems so far, that with lmtx, any standalone file, document, minimal
working example that does not load cmr at the outset does not produce
an acceptable outcome either. By saying acceptable I meant to say it
namely from a typographical point of view. Nothing else.
I have no clue of what you talk about here.
...
from the TeXbook 380-381
«\obeylines doesn’t say ‘\def^^M{\par}’, so we must make any desired changes to
\par before invoking \obeylines. (2) The \uncatcodespecials operation changes a
space to category 12; but the \tt font has the character ‘␣’ in the ⟨space⟩ position, so we
don’t really want ␣12 . (3) The \obeyspaces macro in Appendix B merely changes the
⟨space⟩ character to category 13; active character ␣13 has been defined to be the same
as \space, a macro that expands to ␣10 . This is usually what is desired; for example,
it means that spaces in constructions like ‘\hbox to 10 pt {...}’ won’t cause any
trouble. But in our application it has an undesirable effect, because it produces spaces
that are affected by the space factor. To defeat this feature, it’s necessary either to
say \frenchspacing or to redefine ␣13 to be the same as \␣. The latter alternative is
better, because the former will discard spaces at the beginning of each
line.»
«In theory, this seems like it ought to work; but in practice, it fails in two ways. One
rather obvious failure—at least, it becomes obvious when the macro is tested—is that
all the empty lines of the file are omitted. The reason is that the \par command at the
end of an empty line doesn’t start up a new paragraph, because it occurs in vertical
mode. The other failure is not as obvious, because it occurs much less often: The \tt
fonts contain ligatures for Spanish punctuation, so the sequences ?‘ and !‘ will be
printed as ¿ and ¡ respectively. Both of these defects can be cured by inserting
and
«When INITEX creates a brand new TEX, all characters have a space factor code
of 1000, except that the uppercase letters ‘A’ through ‘Z’ have code 999. (This
slight difference is what makes punctuation act differently after an uppercase letter; do
you see why?) Plain TEX redefines a few of these codes using the \sfcode primitive,
which is similar to \catcode (see Appendix B); for example, the instructions
\sfcode‘)=0
 \sfcode‘.=3000
make right parentheses “transparent” to the space factor, while tripling the stretcha-
bility after periods. The \frenchspacing operation resets \sfcode‘. to
1000.»
Everything you cite above is very likely true for plain TeX, but maybe
not for ConTeXt...
/Mikael
PS I do not think that your emails come out well. In order to get
help, I would suggest a strategy that not so much only sounds as
nagging and complaints. One thing that has been lacking is a clear
explanation of what you really try to achieve.
Mikael, I'll surely heed your advice with that strategy. It's almost
impossible to have a decent looking pdf with lmtx

kerning of a simple word sequence is off, along with spurious spaces
interjected between sentences. What gives?

And with luatex frozen for better or for worse, I better stick to plain
tex

-- 
Thus spake the master programmer:
	"After three days without programming, life becomes meaningless."
		-- Geoffrey James, "The Tao of Programming"