Then the careful writer should not publish it until the errors are
corrected.
But Seebach was the careless writer, and his carelessness, unlike any
carelessness on Schildt's part, caused provable harm to a person.
Seebach labels as errors matters of taste and style. For example,
Seebach claims that the use of upper case on an obviously case-
insensitive file system is an error, and it's not.
Are you going back to your earlier position that it is not possible to
understand a false statement in the real world?
Yes, because understanding is knowledge, and knowledge is justifiable
true belief. I would add that knowledge-of-negatives (such as knowing
that a construct which works on a given platform is not going to work
on another platform because it is "undefined under the standard") is
real knowledge only if you're a Platonist as regards the philosophy of
mathematics.
In the traditional account of philosophy of mathematics, there are
three main schools. Logicism derives from Plato by way of Bertrand
Russell's metaphysical belief that names (such as the "reigning King
of France in 1920") have a referent in all cases, with the "idea" or
"concept" of a Bourbon descendant who is not a pretender substituting
for the real person.
Formalism derives from Hilbert's notion that math is "just" a
meaningless game like Chess or Go, played with symbols.
Constructivism, to which Dijkstra seems heir, is the vastly more
interesting notion that (1) one should not reason from the axiom of
excluded middle (p->(p&~p) therefore ~p) and (2) mathematical entities
(such as computer programs or programming languages) only exist
insofar as they can be sensibly constructed.
In constructivism aka intuitionism, negative knowledge (x is
undefined) is derivative and a verbal translation of positive
knowledge. For example, intuitionism replaces "the square root of -1
is undefined" with a calculus of complex numbers (as does
nonphilosophical math).
Unfortunately, math classes are taught by teachers generally ignorant
about the philosophy of mathematics with the result that a vulgar
Platonism reigns. Thus in the second year of high school there's no
such thing as a complex number, but for the uni math major there is.
This causes Seebach and Heathfield to solemnly and with no little
silliness to say absurd, pernicious and stupid things about C and far
worse, about people. If a C compiler allows the programmer to code
#include <stdio.H> then considered the only sensible way, as human
praxis, C allows this in this instance. If at least one C compiler
evaluates binary ops from left to right in all cases, then C allows
this.
C as a human language follows the same rules as English does in modern
linguistics. Because it was designed before we knew what we were
doing, standardizing C is pissing in the wind.
Most people say, when knocking on a door and asked "who the hell is
it?", "it's me". This is technically a solecism since when you use
"is" and similar verbs of being as opposed to action, you are equating
the subject with the noun phrase on the right hand side of the verb,
therefore the NP on the RHS is a subject, and when it is a pronoun,
must be "I". You're supposed to say "It is I" like some sort of
fruitcake, and this is shown clearly when you add material to the bald
statement: "it is me who knocks on your goddamn door" sounds wrong
whereas "it is I who knock upon your daughter's door" is formally
correct in some grammars (note that the verb must agree with its real
subject, and note that this oddity is less about how to say English
gude and more about the fact that a formal grammar of English is
impossible).
In modern linguistics, since native English speakers who are
reasonably well-educated say "it is me", or "it's me", or "it's me,
goddamnit", the grammar must be revised to accept this.
This means that if the C89 and C99 standards team had no remit to
create a sensible C, and apparently they did not, then at least they
could have done a decent job of describing actual C. This would have
entailed defining named dialects of the language, such as one in which
file identifiers were case insensitive.
But as it is, to say of perfectly sensible constructs and programmer
expectations that they are undefined is like a tedious traditional
linguist telling people to say "it is I". And, the scholastic silly
Standard seems to have only created confusion and harmed reputations.
Most real C programmers are only concerned with what works in their
compiler and OS.
If you don't understand a statement/text/sentence/paragraph/etc, how can you
decide whether it's true, accurate, etc?
If it purports to be a biography of the reigning king of France you do
not in fact know whether it's "really" a novel or a work of "virtual"
history. You have to know the "genre". But Seebach knew the "genre"
and claimed that Herb was clear, which in non-fiction means that he
was understandable, which means he gives knowledge (justified true
belief). All Seebach needed to do was write "Herb writes with an only
apparent clarity" but he did not. The sloppiness infects C: The
Complete Nonsense thoroughly which means it was the pot calling the
kettle black: the use of upper case file identifiers was not an error,
ASCII codes are on most systems in 8 bit units, and most systems use
twos-complement.
