pure scope and try catch scoping idea

[John Rivers]

try catch blocks create a new scope

i don't think that they should

for example, to add error handling to this statement:

int x = 1;

you have to rewrite it as a seperate declaration and assignment:

int x;
try {
x = 1;
} catch {}

maybe if try catch used different braces to show new scope is not
necessary:

try [
//new scope not needed here
int x = 1;
] catch (Exception exc) {
//new scope needed here
}

this would allow error handling to be non-destructively added and
removed from code

plus it should be possible in C# to do this

int x = 1;
{
int y = 1;
}

ie: have a scope for no other purpose than having a scope

rather than

do {
//create new scope for scopes sake
} while (false)

[Christof Nordiek]

Hi John,

comments inlined.

"John Rivers" <fi*****@btinternet.comschrieb im Newsbeitrag
news:11**********************@11g2000cwr.googlegro ups.com...

try catch blocks create a new scope

i don't think that they should

for example, to add error handling to this statement:

int x = 1;

you have to rewrite it as a seperate declaration and assignment:

int x;
try {
x = 1;
} catch {}

maybe if try catch used different braces to show new scope is not
necessary:

try [
//new scope not needed here
int x = 1;
] catch (Exception exc) {
//new scope needed here
}

Even that wouldn't work. If the assignment failed, x is in an unassigned
state. That's why x would be unusable in the catch clause. You could assign
to it. But reading would result in a compiler error: Use of unassigned
variable. So the assignment has to be inside of the catch clause anyway

>
this would allow error handling to be non-destructively added and
removed from code

plus it should be possible in C# to do this

int x = 1;
{
int y = 1;
}

ie: have a scope for no other purpose than having a scope

But this already works in C#. I even used it once.
So there is no use for that:

>
rather than

do {
//create new scope for scopes sake
} while (false)

[Brian Gideon]

John Rivers wrote:

try catch blocks create a new scope

i don't think that they should

for example, to add error handling to this statement:

int x = 1;

you have to rewrite it as a seperate declaration and assignment:

int x;
try {
x = 1;
} catch {}

maybe if try catch used different braces to show new scope is not
necessary:

try [
//new scope not needed here
int x = 1;
] catch (Exception exc) {
//new scope needed here
}

this would allow error handling to be non-destructively added and
removed from code

plus it should be possible in C# to do this

int x = 1;
{
int y = 1;
}

ie: have a scope for no other purpose than having a scope

rather than

do {
//create new scope for scopes sake
} while (false)

John,

Braces can already be used to create "pure scopes". And, I happen to
like that a new scope is created for try-catch-finally blocks. Plus,
it's easy to remember the rule that braces always narrow the scope.

Brian

[John Rivers]

good point re: refering to unassigned variables in catch clause

thanks for pointing out you can already use { } - for some reason I
thought you couldn't
On Jan 12, 4:29 pm, "Christof Nordiek" <[e-mail]:

Hi John,

comments inlined.

"John Rivers" <[e-mail] im Newsbeitragnews:11**********************@11g2000cw r.googlegroups.com...

try catch blocks create a new scope

i don't think that they should

for example, to add error handling to this statement:

int x = 1;

you have to rewrite it as a seperate declaration and assignment:

int x;
try {
x = 1;
} catch {}

maybe if try catch used different braces to show new scope is not
necessary:

try [
//new scope not needed here
int x = 1;
] catch (Exception exc) {
//new scope needed here
}Even that wouldn't work. If the assignment failed, x is in an unassigned

state. That's why x would be unusable in the catch clause. You could assign
to it. But reading would result in a compiler error: Use of unassigned
variable. So the assignment has to be inside of the catch clause anyway

this would allow error handling to be non-destructively added and
removed from code

plus it should be possible in C# to do this

int x = 1;
{
int y = 1;
}

ie: have a scope for no other purpose than having a scopeBut this already works in C#. I even used it once.

So there is no use for that:

rather than

do {
//create new scope for scopes sake
} while (false)- Hide quoted text -- Show quoted text -- Hide quoted text -- Show quoted text -