Fixed string concat to cstring conversion

[wnats]

?str = ("0" ,, "1") ,, "2"

!println(c_string(str))

would incorrectly output "01", at least in my local machine which at that time happened to allocate a zeroed out memory chunk on an lpvm alloc. This problem is more noticeable when calling !logmsg(<log_str>), as <log_str> is first converted into a cstring.

The culprit seems to be the incr(!offset) line. The line is unnecessary as offset has been incremented accordingly in each recursive pack call.

wybe/wybelibs/wybe/string.wybe

Lines 180 to 183 in 35b5c14

	| concat(?left, ?right) ::
	pack(left, !raw, size, !offset)
	pack(right, !raw, size, !offset)
	incr(!offset)

Perhaps unrelated to the main problem, but I have also changed the third argument of mutate to len - 1, which I believe is more accurate, although I have yet to see a corner case for this.

wybe/wybelibs/wybe/string.wybe

Lines 40 to 49 in 35b5c14

	pub def c_string(s:_):c_string = str where {
	if { s = buffer(_, ?str) :: pass
	| else ::
	?len = length(s) + 1
	foreign lpvm alloc(len, ?str)
	foreign lpvm mutate(str, ?str, len, true, len, 0, '\0')
	?offset = 0
	pack(s, !str, len, !offset)
	}
	}

[jimbxb]

LGTM. This is a good catch!

[pschachte]

Thanks for fixing this.