C Syntax

[C# Learner]

Why is C syntax so uneasy on the eye?

In its day, was it _really_ designed by snobby programmers to scare away
potential "n00bs"? If so, and after 50+ years of programming research,
why are programming languages still being designed with C's syntax?

These questions drive me insane. Every waking minute...

[I. Appel]

> Delphi is Pascal with maybe a few bells and whistles.

Therefore, it's based on an older language than C, maybe that's why
it's a bit boring.

Pascal and C are of the same age almost. Pascal maybe 2-3 years older than
C, not more

Well, rephrasing Paul Graham, I should say "Don't wait until C will become as
clear as Python, use Python instead".

If you really think Python has clearer syntax than C, well, think twice.
One terrible fact is this indentation thing in Python to delimit code
blocks. One tab or space is left behind, and the whole thing is screwed.
To me, it's an unacceptable syntax issue.

Python programs are easier to read and understand, that's what I understand
as "clear". And I had never such problems as leaving behind tabs or spaces,
it's generally no problem with good text editor.

[Joe Wright]

Dan Pop wrote:

In <84************ **************@ posting.google. com> ol*****@inspire .net.nz (Old Wolf) writes:

Da*****@cern. ch (Dan Pop) wrote:

None of the programming languages assigning semantics to indentation
has ever become mainstream. There must be a reason...

The language of makefiles?

Indentation per se is meaningless there. It's just that the language
syntax requires a TAB character as the first character on a line in
certain contexts.

Dan

Usually the Rule starts at the left margin and the Command is on the
next line indented with the TAB character. I always found this TAB a
nuisance because my editor puts spaces out when I press Tab key.

Imagine my joy upon reading that GNU make (if not others) also
allows separating Rule from Command with semicolon.

# Use GNU make to build an executable for testing GE.

cc=gcc
obj=ge.o main.o
exe=main.exe
opt=-W -Wall -ansi -pedantic -O2 -c

# Note that semicolon (not TAB) separates Rule from Command on a line.

$(exe) : $(obj) ; $(cc) -s $(obj) -o $(exe)
ge.o : ge.c ; $(cc) $(opt) ge.c
main.o : main.c ge.h ; $(cc) $(opt) main.c

--
Joe Wright mailto:jo****** **@comcast.net
"Everything should be made as simple as possible, but not simpler."
--- Albert Einstein ---

[Goran Larsson]

In article <4t************ ********@comcas t.com>,
Joe Wright <jo********@com cast.net> wrote:

Imagine my joy upon reading that GNU make (if not others) also
allows separating Rule from Command with semicolon.

This is a standard make feature. In a way it is the fundamental
method to separate the dependency list from the commands. The
lines startong with TABs are just continuation lines for the commands.

--
Göran Larsson http://www.mitt-eget.com/

[James Kanze]

C# Learner <cs****@learner .here> writes:

|> Alan Balmer wrote:

|> > On Thu, 27 May 2004 04:36:24 +0100, C# Learner <cs****@learner .here>
|> > wrote:

|> [...]

|> > Forgive me if I'm wrong, but my impression is that you don't have
|> > a deep enough understanding of the C language, and indeed of
|> > programming languages in general, to appreciate the reasons for
|> > the syntax features you're commenting on.

|> I believe that I appreciate the reasons for these features. What I'm
|> saying is that I think that there are better (in terms of allowing
|> the user of the language to write readable code) alternatives that
|> could have been implemented instead.

|> > Take your original example of eliminating the deli meters around
|> > the conditional in an if statement. Think about what this would
|> > mean for compound conditions. Combine that with the Python-style
|> > blocking and then think about an if statement which tests for
|> > several conditions, requiring more than one line to write.

|> a) Valid C syntax:

|> if (foo &&
|> bar) {
|> foobar();
|> }

|> b) Similar code to the above but using my suggested syntax changes:

|> if foo &&
|> bar:
|> foobar();

|> Why wouldn't (b) be feasible here?

|> Everything from 'if' to ':' is considered the condition.

Even if there is a ?: operator in the conditional expression?

|> After the newline after ':', whitespace is required to form a code
|> block.

Let's see, you want to overload on whitespace?

|> > As you study and understand the language, you'll find that it's
|> > all nicely consistent, and that there are good reasons for most of
|> > the features which seem odd to you.

|> The only way in which they seem odd to me is that they make code
|> much less readable than it could be, in my opinion.

Things like C's declaration syntax make code less readable. But you
didn't mention that. The preprocessor can be really effective in making
code less readable too, but you didn't mention that either. And of
course, all those funny characters really made code less readable back
in the days when I had to work with an ISO 646 terminal -- in the German
variant.

But you seem to be attacking areas where C basically has no serious
problems. I personally prefer something like:

if expr then
stmt_list
else
stmt_list
endif

but the alternative used by C is pretty acceptable too. (And my
preference may have been conditioned by the ISO 646 terminals above.
Things like:

if ( expr ) ä
stmt_list
ü else ä
stmt_list
ü

are not particularly readable.)

|> Okay, if you don't agree with the 'if'..':' idea, then how about
|> changing the parentheses required for test conditions for a
|> different pair of characters? An ideal pair would be a pair that
|> isn't used elsewhere in the language, for readability's sake.

|> > One of the characteristics of C is terseness, and extra parens
|> > aren't required by the language for no reason.

|> My point is that a different construct could be substituted in each
|> case.

|> > Also, think about the fact that language inventors and
|> > implementers are, by and large, a pretty bright bunch. In general,
|> > they probably have more and wider experience in the field than you
|> > do, and some of them might even be as smart ;-)

Let's not get carried away. Language designers don't work in a vacuum,
and C is like it is for a variety of historical reasons.

|> So those who invented C's syntax are necessarily brighter than those
|> who invented, say, Python's syntax?

They were certainly working in a different context. I don't know
Python, so I can't say, but from the little you've shown, my impression
is that it is something that works well for quick hacks, but doesn't
scale well.

|> > Those features which have passed through to modern languages have
|> > done so for a reason.

|> I honestly wonder what that reason is.

They work.

--
James Kanze
Conseils en informatique orientée objet/
Beratung in objektorientier ter Datenverarbeitu ng
9 place Sémard, 78210 St.-Cyr-l'École, France [phone]

[James Kanze]

Da*****@cern.ch (Dan Pop) writes:

|> In <c9**********@o ravannahka.hels inki.fi> Joona I Palaste
|> <pa*****@cc.hel sinki.fi> writes:

|> >Mark McIntyre <ma**********@s pamcop.net> scribbled the following:

|> >> Gonads. This is nothing more than a style thing. French people
|> >> put two spaces after a full stop. English people don't. Same
|> >> idea.

|> >I've seen plenty of English-speakers (at least USAns) write two
|> >spaces atfer a full stop.

|> It has nothing to do with the language. It's a common (albeit not
|> universal) typographic convention to leave more space after the
|> punctuation sign terminating a sentence then after punctuation signs
|> inside a sentence.

It is a common (albeit not universal) typographical convention in the
anglosaxon world. Look at a book printed in France or Germany.

Correctly typeset French uses three different spaces :

une espace fine insécable : about a fourth of a quad, no line break
allowed.
une espace mots insécable : about a third of a quad, no line break
allowed.
une espace justifiante : nominally about a third of a quad, but can be
stretched or shrunk, allows a line break.

Nothing larger for the end of a sentence.

|> Examine a few printed books carefully and you may notice it (it's
|> not twice as much, so it's not immediately obvious).

|> The only way to follow this convention in a plain text document is
|> by using two spaces instead of one. This is what I'm consistently
|> doing, no matter in which of the three languages I'm familiar with
|> I'm writing.

Then you're probably doing it wrong in two of them. Different
linguistic communities have different conventions. A larger space after
a sentence is definitly wrong in French or German, and I think it is
wrong in Italian. On the other hand, it would generally be considered
poor typesetting in English not to use it, although how much extra space
is debattable, and if the difference is small enough, the average reader
might miss it.

|> >Now the French, they put a space before an exclamation or question
|> >mark. Like this: "Regardez moi ! Je suis français !". What's
|> >with that, then?

|> Yet another bogus generalisation coming from Joona. Quoting from a
|> post from a French mailing list:

|> ton enfant ? quel age?

|> :-)

Doubtlessly a typo. Or simply incorrect use.

On the other hand, correct typographical use here is an espace fine
insécable, not a full space. Personally, I find the difference
between an espace fine and an espace mots pretty small, and when using a
fixed width font, use a normal espace for both. I know people, however,
who insist on the difference, and drop the espace fine, rather than use
a full space. (They're the exceptions, I think, but I've not done a
statistical analysis to prove it.)

In C, I've noticed that it is immediately obvious when a French speaker
has written the code, even if the comments are in English -- there's a
space in front of the : and the ;, which is never present in code
written by anyone else.

--
James Kanze
Conseils en informatique orientée objet/
Beratung in objektorientier ter Datenverarbeitu ng
9 place Sémard, 78210 St.-Cyr-l'École, France [phone]

[Mark McIntyre]

On 29 May 2004 11:42:44 +0200, in comp.lang.c , James Kanze
<ka***@gabi-soft.fr> wrote:

Da*****@cern.c h (Dan Pop) writes:

|> In <c9**********@o ravannahka.hels inki.fi> Joona I Palaste
|> <pa*****@cc.hel sinki.fi> writes:

|> >Now the French, they put a space before an exclamation or question
|> >mark. Like this: "Regardez moi ! Je suis français !". What's
|> >with that, then?

|> Yet another bogus generalisation coming from Joona. Quoting from a
|> post from a French mailing list:

|> ton enfant ? quel age?

Doubtlessly a typo. Or simply incorrect use.

The second is a typo. I work for a french bank, and the vast majority of
french expats use spaces before and after emphatic punctuation. Only the
anglicised french omit the space.

Remind me what this has to do with C....
--
Mark McIntyre
CLC FAQ <http://www.eskimo.com/~scs/C-faq/top.html>
CLC readme: <http://www.angelfire.c om/ms3/bchambless0/welcome_to_clc. html>
----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! >100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---
----== Posted via Newsfeed.Com - Unlimited-Uncensored-Secure Usenet News==----
http://www.newsfeed.com The #1 Newsgroup Service in the World! >100,000 Newsgroups
---= 19 East/West-Coast Specialized Servers - Total Privacy via Encryption =---

[Viktor Lofgren]

In comp.lang.c C# Learner <cs****@learner .here> wrote:

Alex Moskalyuk wrote:

Have you ever looked into Perl or Lisp code?

C suddenly starts making perfect sense.

What's uneasy to you? Curly braces, parentheses, keywords, or something
else?
One of the biggest flaws in C syntax, in my opinion, is the required
parentheses for test conditions.

Here's a very simple example:

void Foo
{
if (FooBar(Parse(P rocess(GetInput ())))
DoSomething();
}

That is _very_ bad code. What if Parse, Process or GetInput fails,
how would you know what hit you?

Instead,
void Foo(void)
{
type_t a;
if(!(a=get_inpu t()))
puts("GetInput died");
if(!(a=process( a)))
puts("process died");
if(!(a=parse(pa rse())
puts("Parse died");
if(foobar(a)
do_something();
}
Imagine if, instead, we could write the following:

void Foo
{
if FooBar(Parse(Pr ocess(GetInput( ))):
DoSomething();
}
This is why: it would not be consistant with the C code-style. The
style is in fact is very obvious when properly idented and highligthed,
Another nicety about Python is the fact that whitespace is used for
defining code blocks. This makes code much clearer than the equivalent
C code, which requires block being/end markers.
I don't like the idea of having indentation as a part of the syntax,
very hard to write longer code on 80x25 terminals :-(
Even if one's accustomed to C syntax, the former is still clearer and
easier-to-read, don't you think?
Programming languages are not meant to be easy to read.
If you want easy-to-read, comment your code.
Why C? Why?!

Here's why: It works and is perfectly balanced between levels. It is not
gibberish like LISP or Assembly (no offense, but it is kinda hard to see what
(+ 2 (* 8 (expt 25 (/ 5 2)))) does). And you still consistent (unlike BASIC)
and doesent require you to write a smaller novel (unlike COBOL) to get your
app to do something. Python may be an OK language, but it still doesent
have the maturity of C.

--
Viktor Lofgren, Umea, Sweden
Mainly self-eductated UNIX programmer, running Slackware-current.

[CBFalconer]

Viktor Lofgren wrote:

Learner <cs****@learner .here> wrote:

Alex Moskalyuk wrote:

Have you ever looked into Perl or Lisp code?

C suddenly starts making perfect sense.

What's uneasy to you? Curly braces, parentheses, keywords, or
something else?

One of the biggest flaws in C syntax, in my opinion, is the
required parentheses for test conditions.

Here's a very simple example:

void Foo
{
if (FooBar(Parse(P rocess(GetInput ())))
DoSomething();
}

That is _very_ bad code. What if Parse, Process or GetInput
fails, how would you know what hit you?

Instead,
void Foo(void)
{
type_t a;
if(!(a=get_inpu t()))
puts("GetInput died");
if(!(a=process( a)))
puts("process died");
if(!(a=parse(pa rse())
puts("Parse died");
if(foobar(a)
do_something();
}

That is still foul and faulty code. Assuming there is an error
value 0, try:

void Foo(void)
{
type_t a;

if (!(a = get_input())) puts("GetInput died");
else if (!(a = process(a))) puts("process died");
else if (!(a = parse(a()) puts("Parse died");
else if (foobar(a) do_something();
/* else donothing */
}

If there is no such error value, the original is much clearer.

--
Chuck F (cb********@yah oo.com) (cb********@wor ldnet.att.net)
Available for consulting/temporary embedded and systems.
<http://cbfalconer.home .att.net> USE worldnet address!

[Arthur J. O'Dwyer]

On Sat, 29 May 2004, Viktor Lofgren wrote:

In comp.lang.c C# Learner <cs****@learner .here> wrote:

if (FooBar(Parse(P rocess(GetInput ())))
DoSomething();
That is _very_ bad code. What if Parse, Process or GetInput fails,
how would you know what hit you?

<on-topic for once> This is the general argument in favor of
writing fail-immune functions. For example, if 'GetInput' fails
it will (let's say) return NULL. Now, 'Process' knows that it
will be very tempting for the client to write 'Process(GetInp ut())';
thus, 'Process' codes as follows:

void *Process(const void *input)
{
if (input == NULL) return NULL;
[...]
}

Now 'Process' can handle anything 'GetInput' throws at it, and it
can do it in a reasonable manner. Apply the same solution to
'Parse' and 'FooBar'. Now the client can safely write

if (FooBar(Parse(P rocess(GetInput ())))
DoSomething();
else
/* Something went wrong in one of those four functions */
;

This is a nice idiom to learn and use; unfortunately, the C
standard library doesn't use it in most places. Especially in
'strcpy' it would be convenient, so that

#define stdup(s) (strcpy(malloc( strlen(s)+1), (s)))

wouldn't exhibit undefined behavior if 'malloc' were to fail,
but sensibly return NULL.
[Not that I'm advocating a change to 'strcpy'; just pointing
out one of the disadvantages of a model that *doesn't* use the
fail-immune handling shown above.]
<back off-topic>
I don't like the idea of having indentation as a part of the syntax,
very hard to write longer code on 80x25 terminals :-(

The canonical solution applies here: If your lines are too long
to be read comfortably by the average human, you ought to be writing
shorter lines. And/or smaller functions with fewer levels of
indentation.

Even if one's accustomed to C syntax, the former is still clearer and
easier-to-read, don't you think?

Programming languages are not meant to be easy to read.
If you want easy-to-read, comment your code.

Don't be ridiculous! Languages are *by definition* meant to make
communication simpler. And until computers learn to talk properly,
that communication is only going to happen through reading. C's
syntax, like that of every (non-joke) language in existence, was
designed to make reading and writing as easy, pleasurable and
effective as possible. The fact that some people like to write
impenetrable C code is no stranger than that some people like to
write impenetrable English or French.

-Arthur

[Dr Chaos]

On Wed, 26 May 2004 22:02:42 +0100, C# Learner <cs****@learner .here> wrote:

Why is C syntax so uneasy on the eye?

In its day, was it _really_ designed by snobby programmers to scare away
potential "n00bs"?
No, but the programmers who designed it didn't care about "n00bs";
their target audience was people who worked at Bell Labs. They were
all very very smart. And it was in the days where actually printing
programs and typing them in was physically expensive and difficult
(literal paper-oriented terminals and tty inputs). So severely
limiting characters was a desirable major goal then. It isn't now.
If so, and after 50+ years of programming research,
why are programming languages still being designed with C's syntax?
Profound anti-intellectualism among practitioners, and cargo-cult
imitation. This spurious idea being that "gee, C was popular; so why
don't we make our language with Cish syntax so we might be popular
too."

Yes, these things have been tested empirically (though that field
is woefully underfunded) and yes the answer comes back that other
syntaxes are less error prone.

In particular the Fortran 90 and Ada typical syntaxes come out on
top.
These questions drive me insane. Every waking minute...

The answers are easy. But depressing.