• Sean Kelly (15/15) Jun 08 2005 The old horse still has a bit of life in it!
• Matthew (2/18) Jun 10 2005
• Unknown W. Brackets (4/32) Jun 11 2005 If I do so recall, I posted about why I still think this is necessary,
• Ben Hinkle (24/43) Jun 13 2005 Once Exceptions get stack trace abilities the file/lineno will be less

Sean Kelly <sean f4.ca> writes:

The old horse still has a bit of life in it!

If we can assume that the current assert expression to evaluate as:



Is there any chance this could be changed to:



Since the default behavior of assert in D is to throw an exception, it seems
quite reasonable that the programmer be given a means to supply the message that
exception carries.  For DMD, this would change:



to:



I'd do this myself except assert is treated as a keyword in D, and it seems
unreasonable to ask users to call a different function if they want to supply a
message.


Sean

"Matthew" <admin.hat stlsoft.dot.org> writes:

gets my vote

"Sean Kelly" <sean f4.ca> wrote in message
news:d87tsv$2sd9$1 digitaldaemon.com...
 The old horse still has a bit of life in it!

 If we can assume that the current assert expression to evaluate as:



 Is there any chance this could be changed to:



 Since the default behavior of assert in D is to throw an exception, it seems
 quite reasonable that the programmer be given a means to supply the message
that
 exception carries.  For DMD, this would change:



 to:



 I'd do this myself except assert is treated as a keyword in D, and it seems
 unreasonable to ask users to call a different function if they want to supply a
 message.


 Sean

 

"Unknown W. Brackets" <unknown simplemachines.org> writes:

If I do so recall, I posted about why I still think this is necessary, 
despite the response that file and line numbers should be enough for 
developers (an argument that obviously ignored open source software.)

-[Unknown]


 The old horse still has a bit of life in it!
 
 If we can assume that the current assert expression to evaluate as:
 

 
 Is there any chance this could be changed to:
 

 
 Since the default behavior of assert in D is to throw an exception, it seems
 quite reasonable that the programmer be given a means to supply the message
that
 exception carries.  For DMD, this would change:
 

 
 to:
 

 
 I'd do this myself except assert is treated as a keyword in D, and it seems
 unreasonable to ask users to call a different function if they want to supply a
 message.
 
 
 Sean
 
 

"Ben Hinkle" <bhinkle mathworks.com> writes:

"Sean Kelly" <sean f4.ca> wrote in message 
news:d87tsv$2sd9$1 digitaldaemon.com...
 The old horse still has a bit of life in it!

 If we can assume that the current assert expression to evaluate as:



 Is there any chance this could be changed to:



 Since the default behavior of assert in D is to throw an exception, it 
 seems
 quite reasonable that the programmer be given a means to supply the 
 message that
 exception carries.  For DMD, this would change:



 to:



 I'd do this myself except assert is treated as a keyword in D, and it 
 seems
 unreasonable to ask users to call a different function if they want to 
 supply a
 message.


 Sean

Once Exceptions get stack trace abilities the file/lineno will be less 
crucial (though the linno will still let you tell which assert in a given 
function failed). If one doesn't mind writing a bit more for message asserts 
here's a simple solution
private import std.format;
void assertStr(int val,...) {
    char[] s;
    void putc(dchar c){ std.utf.encode(s, c); }
    if (!val) {
        // I would call std.string.sformat but it takes ... so
        // I can't :-(   Someone should add va_list versions.
        std.format.doFormat(&putc, _arguments, _argptr);
        throw new Exception(s);
    }

}
int main() {
    int a,b;
    a = 10;
    version(Assert) assertStr(a == b, "oops, b was %d",b);
    version(Assert) assertStr(a == b, "%s(%d) another",__FILE__,__LINE__);
    return 0;
}