Brent Dax wrote:
Dan Sugalski wrote on 10/27/02 8.11:
I tried very hard to make sure that there was always a valid
interpreter. --
When I was working on switching most fprintf calls to PIO, there were so
many functions that didn't take an interpreter that I eventually made
PIO_printf and PIO_eprintf (output to stderr) fall back to stdio if
given a null interpreter.
The problem is the Parrot_vsprintf_c call prior to output. When we limit the format/args to fprintf(3) conventions, when there is no interpreter, we could use fprintf as a fallback.
But actually, the only case, when there is no interpreter, is when constructing the interpreter itself fails.
... So at least part of the problem is that lots of functions don't even have an interpreter available.
I'll walk through them and the interpreter startup code and change functions to take an interpreter argument.
leo
