Re: [Caml-list] Expanding the Float Array tag

[Goswin von Brederlow <goswin-v-b@web.de>]

On Thu, Sep 19, 2013 at 06:18:38AM -0000, oleg@okmij.org wrote:
> 
> I am curious about the focus on generalizing double_tag or
> double_array_tag when there is already a seemingly good candidate for
> tagging values that should not be scanned. I mean the
> string_tag. Please search the OCaml code for string_tag and
> String_tag. You see it is mentioned in generic routines like
> generic comparison compare.c -- which uses memcpy anyway, thus
> treating the string just like a sequence of bytes. Other generic
> functions like hash are similar. The only other non-trivial use of
> string_tag is in printexc and the likes. Those cases are indeed
> problematic. Other than them, string_tag can denote arbitrary opaque
> byte array and noone will notice. Incidentally, if by string we mean
> UTF-8 or UTF-16 string, it is essentially a sequence of bytes anyway.

Strings are special in that they have a length in bytes. The length in
words is taken from the header, multiplied by the word size and the
delta to the real string length is stored in the last byte of the block.

Reusing the string tag would confuse code that relies on this, like
marshaling.

MfG
	Goswin