I think that is the wrong way to go about things. I think it is better to make a more_human_readable_werror() that operates differently on %O, and use that one when presenting backtraces and hilfe-data.

From the Pike manual:

*! @value 'O' *! Any value, debug style. Do not rely on the exact formatting; *! how the result looks can vary depending on locale, phase of *! the moon or anything else the _sprintf method implementor *! wanted for debugging.

"Do not rely on the exact formatting"!

Exactly what kind of code does people have that depends on %O output format? I use %O quite a bit, but only for printing debug info to stdout or similar.

Apparently there is no such consensus.

That's still true.

For instance Pike,

array_sscanf(""foo" "bar"","%O");

(1) Result: ({ /* 1 element */ "foo" })

(Which *I* think is the correct behavious in this case and sprintf is broken.)

then we maybe need something else that allows to output a mixed variable without having to check the type first.

often i want to print strings without having to worry if they are initialized, and i either use %O or %s with (string)foo, to make sure the 0 is casted to " " so that %s won't complain.

occaionally i want to print numbers without having to worry about the type. or what do i do if the variable is mixed?

currently %O is the only operator that can handle those.

maybe something like %M or %m would be nice, or, even better, some notation that reflects the types we have:

string|int foo; sprintf("%s|%i)", foo); // the above is not a good notation for this, but maybe this is: sprintf("%(si)", foo);

greetings, martin.