From Newsgroup: comp.lang.c

On 12/12/24 06:38, Janis Papanagnou wrote:

On 12.12.2024 06:35, Keith Thompson wrote:

Janis Papanagnou <janis_papanagnou+ng@hotmail.com> writes:
[...]

For (yet another) example; my K&R shows a syntax for expressions like

expression := binary
binary := binary + binary
binary := binary * binary

That's odd. Is that an exact quotation?

No, not exact, I abbreviated it; omitted about 25 other operators and
used another syntax (no ' | '). (Trust me that it hasn't things that
I called 'factor' and 'term' in my post, which is equivalent to what
you have formulated below in your copy.) - I'm using a translation of something that someone classified as being a "pre-second" edition but
not quite the first edition. Two references point to 1977 (Prentice
Hall) and 1978 (Bell Labs). - The text for the "binary" syntax has two optional informal columns, the first one has the comment "precedence"
for some of the variants of "binary" operators. (But it's also just
titled as "Syntax in Short"; probably presented in a form to make it
easy to understand without overloading it for purpose of a textbook.)
It serves the purpose to explain an ambiguous syntax with non-codified precedence and a separate precedence table (but it's not an exact "C"
syntax description as you'd probably find it in standards documents).

My copy of K&R 1st edition has nothing remotely resembling that. I
cannot find "binary" as an element of the grammar anywhere. There are 16 grammar rules for "expression". The one that comes closest is
"expression binop expression", where binop is one of C's 19 binary
operators, divided into 12 different priority levels.

--- Synchronet 3.20a-Linux NewsLink 1.114

From Newsgroup: comp.lang.c

On 18.12.2024 22:04, James Kuyper wrote:

On 12/12/24 06:38, Janis Papanagnou wrote:

[...]

My copy of K&R 1st edition has nothing remotely resembling that.

I think someone here already mentioned that the English version
looks different compared to the translation I've in my bookshelf.
(So I am not sure what your post is actually intending.[*])

(But that's anyway unimportant to the point I made; that you can
define semantical information like the precedence separately or
by syntax.[**])

Janis

PS: FYI; Some of your posts, James, arrive also in my mailbox.

[*] I've put (if you're interested) a scan of that page uploaded
here: http://volatile.gridbug.de/KR_syntax-rotated90.pdf
(But similar syntaxes for expressions can be found also in other
programming languages' contexts; see below.)

[**] There's various levels of abstraction possible. For example
- describe expressions in- or semi-formally (e.g. in Bolsky/Korn)
- describe them without precedence semantics ("K&R DE version")
- reflect operator precedence in syntax (quite typical)
- allow extensions/change of operators and precedence (Algol 68)
(Sometimes there's no clear line; e.g. part of the definitions
for expressions in Bolsky/Korn are like in the "K&R DE version",
and the precedence is defined by the order they are listed.)

[...]

--- Synchronet 3.20a-Linux NewsLink 1.114

From Newsgroup: comp.lang.c

On 12/18/24 19:35, Janis Papanagnou wrote:

On 18.12.2024 22:04, James Kuyper wrote:

On 12/12/24 06:38, Janis Papanagnou wrote:

[...]

My copy of K&R 1st edition has nothing remotely resembling that.

I think someone here already mentioned that the English version
looks different compared to the translation I've in my bookshelf.
(So I am not sure what your post is actually intending.[*])> (But that's anyway unimportant to the point I made; that you can
define semantical information like the precedence separately or
by syntax.[**])

Janis

PS: FYI; Some of your posts, James, arrive also in my mailbox.

Yes, as I've explained before, in April, 2021, Thunderbird changed from
having a Reply button that sends the response to the newsgroup, to
having a Followup button that does the same thing, and a Reply button
that sends the response by e-mail to the author. I spent a couple of
decades learning to hit the Reply button, and at 66 years of age, after
three years, I'm still finding it difficult to unlearn that habit,
though I've made some progress.

[*] I've put (if you're interested) a scan of that page uploaded
here: http://volatile.gridbug.de/KR_syntax-rotated90.pdf
(But similar syntaxes for expressions can be found also in other
programming languages' contexts; see below.)

It's not just that it "looks different". It is substantially different
from the First Edition, which had "expression binop expression" as one
of the rules for expression, with a precedence table for binop. It is
also substantially different from the second edition, which closely
matched C89, with a separate grammar production for each level of
precedence. Nor does it look like a plausible intermediate step between
those formulations, which is roughly how you described it.
--- Synchronet 3.20a-Linux NewsLink 1.114

From Newsgroup: comp.lang.c

James Kuyper <jameskuyper@alumni.caltech.edu> wrote:

On 12/12/24 06:38, Janis Papanagnou wrote:

On 12.12.2024 06:35, Keith Thompson wrote:

Janis Papanagnou <janis_papanagnou+ng@hotmail.com> writes:
[...]

For (yet another) example; my K&R shows a syntax for expressions like

expression := binary
binary := binary + binary
binary := binary * binary

That's odd. Is that an exact quotation?

No, not exact, I abbreviated it; omitted about 25 other operators and
used another syntax (no ' | '). (Trust me that it hasn't things that
I called 'factor' and 'term' in my post, which is equivalent to what
you have formulated below in your copy.) - I'm using a translation of
something that someone classified as being a "pre-second" edition but
not quite the first edition. Two references point to 1977 (Prentice
Hall) and 1978 (Bell Labs). - The text for the "binary" syntax has two
optional informal columns, the first one has the comment "precedence"
for some of the variants of "binary" operators. (But it's also just
titled as "Syntax in Short"; probably presented in a form to make it
easy to understand without overloading it for purpose of a textbook.)
It serves the purpose to explain an ambiguous syntax with non-codified
precedence and a separate precedence table (but it's not an exact "C"
syntax description as you'd probably find it in standards documents).

My copy of K&R 1st edition has nothing remotely resembling that. I
cannot find "binary" as an element of the grammar anywhere. There are 16 grammar rules for "expression". The one that comes closest is
"expression binop expression", where binop is one of C's 19 binary
operators, divided into 12 different priority levels.

Apparently translators took freedom to make various changes.
Polish translation of 1st edition has rules with translated
names for nontermials. There are 17 rules for expressions,
one of which contains binary operator. Text following this
listens 18 binary operators, divided into 10 priority levels.
Translators claimed that they removed inaccuracies, but
probably planted extra confusion by having extra rule for
comma operator and relating priority of assignment operators
and other operators.

BTW: Polish translators also changed examples for K&R style
to Allman style...
--
Waldek Hebisch
--- Synchronet 3.20a-Linux NewsLink 1.114

From Newsgroup: comp.lang.c

Michael S <already5chosen@yahoo.com> writes:

On Wed, 11 Dec 2024 21:19:54 +0000
bart <bc@freeuk.com> wrote:

This also comes up with 'while (cond); {...}'.

$ cat foo.c
void foo(int x)
{
while (x--);
bar();
}

$ clang-format < foo.c
void foo(int x) {
while (x--)
;
bar();
}

Do I use clamg-format myself? No, I don't.

And I don't know why C formatters are generally less popular among C programmers then, for example, go formater among Go programmers or Rust formatter among Rust programmers.

I don't have any experience with Go formatters or Rust formatters.
My experience with C formatters is that they format too much; in
some ways they help, but in other ways they interfere with choices
that are incidental to the high-level adjustments desired. It
would be nice to see a code formatter that respects low-level
choices while adjusting high-level format rules. To say this
another way, the formatting that is done is too rigid - the tool
always adjusts everything, instead of adjusting only those aspects
that are needed to fix the high-level transgressions.
--- Synchronet 3.20a-Linux NewsLink 1.114

From Newsgroup: comp.lang.c

Keith Thompson <Keith.S.Thompson+u@gmail.com> writes:

Tim Rentsch <tr.17687@z991.linuxsc.com> writes:

Keith Thompson <Keith.S.Thompson+u@gmail.com> writes:

If someone wanted to ensure that all static functions defined in a
translation unit are declared near the top, there could be a separate
tool to generate, or at least check, the declarations. I'm not aware of >>> any such tool, which suggests there probably isn't much demand for it.

What it suggests to me is that there are tools being used that
you aren't aware of.

What that suggests to me is that you feel the need to snipe without
sharing any actual information.

I'm not sniping. I'm just making a statement about what the stated
facts bring to mind, in my mind. You made a similar kind of
statement. Is there some reason you think you should be entitled
to do that and I should not be?
--- Synchronet 3.20c-Linux NewsLink 1.2