portably.

Also, we're really interested in termios stuff, so we don't actually want
to call setvbuf anyway.

controlling the vbuf of a stream.

when it fails (neither uudecode nor ydecode can fail currently).

File.iniGetValue() method returned a garbage string if the value requested
did not exist in the ini section and the default value specified was a string:
cstr2 will point to cstr in this case, so we can't free(cstr) before copying
cstr2 into a new JSVAL.

don't read/write the file for *every* property!
Fixed several apparent memory leaks (upon errors) and at least one
freed-pointer usage.

or 'longs'. Amazing that no one noticed.

MS's CRTL throws an assertion. Buffer size must be >=2.

cleaning up the open mode rather than truncating at the first 'e' char.  Also,
allow passing more than three flags through to fdopen().

successfully opened.

NULL because the get function is called on the prototype which isn't
constructed (via the constructor).

resolve callbacks.

all exported functions and data types using a 32-bit time_t (time32_t).

When a minimum number of parameters is specified in the method spec, this
shouldn't be required, but overkill is better than underkill.

NOTE: argv[argc] != JSVAL_VOID anymore.

Only js_ValueToStringBytes() still leaks.

(Everything still broken)

      on the floor of Deuce's computer room
Cyan: Rob's on his way
* Deuce throws his coffee on it.

File.iniGetObject(): If the requested section didn't exist (in the .ini file),
the script would terminate (rather than undefined being returned).

if the specified section doesn't exist (or contains no keys/values).

This should fix Matt's problem with crashes when calling File.writeAll() with a
null value and other potential occurances of the same bug.

unsupported mode char in it (e.g. 'e'), i.e. from load/lockfile.js.

SpiderMonkey actually use rval.