Why was there no compiler error?

[Man-wai Chang to The Door (24000bps)]

http://www.kernel.org/diff/diffview.cgi?file=/pub/linux/kernel/v2.6/incr/patch-2.6.32.3-4.bz2;z=112

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.32.3
^ ^ 17:13:02 up 4 days 6:24 1 user load average: 3.33 2.68 1.79
ä¸Â借貸! ä¸Âè©Â騙! ä¸Âæ´交! ä¸Â打交! ä¸Â打劫!
ä¸Â自殺! 請考慮綜æ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa

 

[Bart van Ingen Schenau]

http://www.kernel.org/diff/diffview.cgi?file=/pub/linux/kernel%....

The link shows a code change from code along the lines of:
a = 42, foo(a);
to
a = 42;
foo(a);

The question, as placed in the subject is: Why did the compiler not
complain about the old code?

The answer is that the old code uses the comma-operator to separate
two expressions.
The rewrite is semantically identical to the original code. The
rewrite is preferable, because it is mode ideomatic code and leads to
fewer questions.

Bart v Ingen Schenau

 

[Man-wai Chang to The Door (24000bps)]

The link shows a code change from code along the lines of:

a = 42, foo(a);
to
a = 42;
foo(a);
The rewrite is semantically identical to the original code. The
rewrite is preferable, because it is mode ideomatic code and leads to
fewer questions.

Thanks

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.32.3
^ ^ 19:59:02 up 4 days 9:10 1 user load average: 3.62 3.52 3.39
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa

 

[Timothy Madden]

http://www.kernel.org/diff/diffview.cgi?file=/pub/linux/kernel/v2.6/incr/patch-2.6.32.3-4.bz2;z=112

Because the comma (",") is also an operator in C++, which can be used in
expressions, not only a punctuator. So the following is a valid C++
expresion:

_expr_, _expr_, _expr_, ...

sama as (a + b) or (expr + expr + expr + ...) is a valid expression,
and, as any expression, it can be used in an expression-statement

_expr_, _expr_ ;
a + b ;

by adding the semicolon ";".

So the old code

request->alpha2[0] = alpha2[0];
request->alpha2[1] = alpha2[1];
request->initiator = NL80211_REGDOM_SET_BY_USER,

queue_regulatory_request(request);

was seen as

request->alpha2[0] = alpha2[0];
request->alpha2[1] = alpha2[1];

( request->initiator = NL80211_REGDOM_SET_BY_USER ) ,
( queue_regulatory_request(request) );

which is valid C++, with the last two lines being an
expression-statement, and which fortunately does the same thing as the
new patched version.

Timothy Madden

 

[Man-wai Chang to The Door (24000bps)]

which is valid C++, with the last two lines being an

expression-statement, and which fortunately does the same thing as the
new patched version.

1. smells like ambiguity though harmless, doesn't it?
2. really no side-effect in terms of compilation?
3. did that kernel programmer want to make a point?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.32.4
^ ^ 21:55:01 up 4:58 0 users load average: 1.00 1.02 1.00
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa

 

[Timothy Madden]

1. smells like ambiguity though harmless, doesn't it?
2. really no side-effect in terms of compilation?
3. did that kernel programmer want to make a point?

No side-effects.

When you know about it since a long time, the ambiguity is only in the
potential for someone else reading your code to be confused, which is
why the patched (corected) version should be used.


Looking at the blank line between the sub-expressions of the
comma-expression in the last statement in the old code, one could say
there was no point to be made by the kernel programmer, just a typo...

 

[Man-wai Chang to The Door (24000bps)]

int a = 42, foo(a);

then foo(a) is not called any more. Instead, a new variable foo is created
and initialized by the value of a. One might not notice this change because
the code still compiles, but of course it does not work properly any more
when used by the most important customer. So I would say the code is a bit
fragile.

Now that's horrible...

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.32.4
^ ^ 16:51:02 up 23:54 0 users load average: 1.00 1.00 1.00
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa