Positioning of catch(...)

[Taras_96]

Hi everyone,

I recently came across a quiz question. Does the standard say anything
about catch(...) being the first catch statement?

try
{
}
catch(...) // <- 'hides' subsequent catching
{
}
catch(A)
{
}
catch(B)
{
}

I've seen numerous examples where catch(...) appears first (eg:
Bjorne), but code doesn't compile using MS C++, so does this mean that
the standard allows it, but some compilers don't?

Taras

 

[Kai-Uwe Bux]

Hi everyone,

I recently came across a quiz question. Does the standard say anything
about catch(...) being the first catch statement?

Yes, catch statements are tried in order [15.3/5] and if present, a
catch(...) shall go last.

try
{
}
catch(...) // <- 'hides' subsequent catching
{
}
catch(A)
{
}
catch(B)
{
}

I've seen numerous examples where catch(...) appears first (eg:
Bjorne), but code doesn't compile using MS C++, so does this mean that
the standard allows it, but some compilers don't?

The standard says [15.3/6]:

A ... in a handler?s exception-declaration functions similarly to ... in a
function parameter declaration; it specifies a match for any exception. If
present, a ... handler shall be the last handler for its try block.

That says that a program where catch(...) is not the last handler in a block
is ill-formed ("shall").


Best

Kai-Uwe Bux

 

[Pranav]

Hi everyone,

I recently came across a quiz question. Does the standard say anything
about catch(...) being the first catch statement?

Yes, catch statements are tried in order [15.3/5] and if present, a
catch(...) shall go last.

try
{
}
catch(...) // <- 'hides' subsequent catching
{
}
catch(A)
{
}
catch(B)
{
}

I've seen numerous examples where catch(...) appears first (eg:
Bjorne), but code doesn't compile using MS C++, so does this mean that
the standard allows it, but some compilers don't?

The standard says [15.3/6]:

  A ... in a handler?s exception-declaration functions similarly to ... in a
  function parameter declaration; it specifies a match for any exception. If
  present, a ... handler shall be the last handler for its try block.

That says that a program where catch(...) is not the last handler in a block
is ill-formed ("shall").

Best

Kai-Uwe Bux

From Wheer can I get C++ standard or it is just Bjarne Strostrup CPP
book..,

 

[Juha Nieminen]

That says that a program where catch(...) is not the last handler in a block
is ill-formed ("shall").

What does "ill-formed" mean in practice? Does it have something to do
with "undefined behavior"?

 

[Erik Wikström]

What does "ill-formed" mean in practice? Does it have something to do
with "undefined behavior"?

I believe that ill-formed code is not required to compile, while
undefined behaviour refers to the run-time behaviour (which means that
the code did compile).

 

[James Kanze]

What does "ill-formed" mean in practice? Does it have
something to do with "undefined behavior"?

[/QUOTE]

I believe that ill-formed code is not required to compile,
while undefined behaviour refers to the run-time behaviour
(which means that the code did compile).

Sort of. Ill-formed means that the compiler must issue a
diagnostic. Having done so, it is free to compile the code if
it wishes. (Good compilers generally don't.) Undefined
behavior means just that---whatever the compiler does with the
code is correct as far as the standard is concerned. It may
refuse to compile it (with or without an error message); it may
compile it do something that works; or it may compile it to do
anything else.