You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Dean Gaudet <dg...@arctic.org> on 1997/07/09 02:32:06 UTC
Re: [PATCH] PR#344: 64-bit cleanup
On Tue, 8 Jul 1997, Alexei Kosut wrote:
> On Tue, 8 Jul 1997, Dean Gaudet wrote:
>
> > Introduce a new type generic_data which is large enough to hold a pointer
> > or an integer, and which the compiler is likely to pass by value
> > efficiently. This requires certain functions that take "void *" to start
> > taking generic_data. It also changes the command_rec and related
> > structures to include a generic_data field instead of a void * field.
>
> Is this really neccessary? It seems a bit drastic.
Theoretically it is necessary to dictate this. For example in the future
there might be a box with 64-bit pointers and 128-bit integers... for what
reason I can't imagine. Just as there are boxes now with 64-bit pointers
and 32-bit integers. So you can't say if the largest "register" type is a
pointer or an integer.
We want the type to be held in a register because otherwise it imposes a
bunch of function-call overhead. A struct or union is (generally) passed
by reference-to-copy rather than being passed by value. unions can only
have their first element initialized at compile time, making them more
inconvenient.
> > The command_rec change required changes to every module to deal with the
> > static initializer for the info field.
>
> Hmm. Does it actually require the change, or is it of the "make it, or
> you'll get a warning, but it'll work okay on 32-bit systems" variety? If
> the former, I'd be very hesitant to agree to the patch unless there are
> absolutely no other solutions.
make it or get a warning, but it'll work okay.
This is one of those things where we just have to bite it and do it. We
hashed all this out the last time I brought up PR#344 I thought.
Dean
Re: [PATCH] PR#344: 64-bit cleanup
Posted by Alexei Kosut <ak...@organic.com>.
On Tue, 8 Jul 1997, Dean Gaudet wrote:
[Dean's 64-bit patch]
> make it or get a warning, but it'll work okay.
>
> This is one of those things where we just have to bite it and do it. We
> hashed all this out the last time I brought up PR#344 I thought.
Yeah, I just wasn't paying as much attention as I should have been.
At any rate, I guess we have to do this. Haven't had a chance to look
hard at the patch, but +1 on the concept (unless there's a better one
*grin*).
-- Alexei Kosut <ak...@organic.com>