[omniORB] proposed patch for Any handling of octet sequences

[David Riddoch]

Hi,


M'self and Wez (jnw@uk.research.att.com) have coded up a performance
optimisation for insertion/extraction of sequences of simple types to/from
Anys (octet, boolean, char, short, ushort, long, ulong, float, double).

This is only available for omniORB 3pre -- since it makes incompatible
changes to the stub/library interface.  Get it from CVS tomorrow as
usual...

I haven't had a change to test it fully, so I'd appreciate it if someone
could use it in anger and let me know if it goes okay.


Cheers,
David



On Mon, 3 Jan 2000, Renzo Tomaselli wrote:

>     I attached a proposed patch for tcParser.cc (related to ver. 2.8). While
> being simple, this patch would quite speed-up insertion/extraction of octet
> sequences to/from anys (see my previous message on this issue a few weeks
> ago).
> The way it works now is definitely a no-go (e.g. handling such sequences
> byte-by-byte) for graphics applications or other applications needing to
> handle large byte blocks through anys.
> A similar patch should be applied to tcParser::calculateItemSize() and
> tcParser::appendItem() as well.