Mailing-List: contact test-dev-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: test-dev@httpd.apache.org
Date: Tue, 17 Dec 2002 13:22:05 -0800
Subject: Re: relative_times_report_t
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v548)
From: Aaron Bannert <aaron@clove.org>
To: test-dev@httpd.apache.org
Content-Transfer-Encoding: 7bit
In-Reply-To: <3DFF893C.4090403@zk3.dec.com>
Message-Id: <976C4FF6-1205-11D7-9902-000393B3C494@clove.org>


On Tuesday, December 17, 2002, at 12:29  PM, David Hill wrote:

>
> Hi all,
>
> Two problems, and a couple of minor issues encountered trying to build 
> flood from CVS on tru64:
>
> first: in flood_report_relative_times.c we have:
>
> typedef void relative_times_report_t;
>
> apr_status_t relative_times_report_init(report_t **report, config_t 
> *config,
>                              const char *profile_name, apr_pool_t 
> *pool)
> {
>    *report = apr_palloc(pool, sizeof(relative_times_report_t));
>
>
> Our compiler (I would think correctly) chokes on the sizeof(void) in 
> the apr_palloc().
> What confuses me is what this what "report" is used for. Looks to me 
> as if it is a placeholder and is not really used. If that is indeed 
> the case then switching to something like void* would make my compiler 
> happy. On the other hand.... switching it to use:
>    *report = apr_palloc(pool, sizeof(report_t *));
> seems just as logical to me, as that matches the return type better.

It's probably just a typo, so doing sizeof(report_t*) is probably the 
right
way to go.

> second: Tru64 does not have (nor need) a strtoll, our long is 64 bit 
> anyway. Nor do we have strtoq (whatever that does). The following 
> fixes it for me:
>
> ddhill@ajo:/home/ddhill/apache/flood/httpd-test/flood
> # ~/bin/diff -u config.h.orig config.h
> --- config.h.orig       2002-12-17 13:36:33.000000000 -0500
> +++ config.h    2002-12-17 13:51:25.000000000 -0500
> @@ -80,6 +80,8 @@
> #endif
> #elif !FLOOD_HAS_STRTOLL && FLOOD_HAS_STRTOQ
> #define strtoll(p, e, b) strtoq(p, e, b)
> +#else
> +#define strtoll(p, e, b) strtol(p, e, b)
> #endif
> #endif  /* __config_h */
>
>
> It could be argued I suppose that this should be special cased for 
> Tru64 using
>
> #if defined(__alpha) && defined(__osf__)
>
> but I would think that in the absence of strtoll or strtoq, this is 
> the best fallback.

That might be an ok workaround, but we should eventually
hide this inside APR, since this will probably bite us later.

> A minor (probably issue): in flood_net.h and flood_net_ssl.h, for the 
> function read_socket(),
> buflen should really be a apr_size_t. On Tru64, this is a long and not 
> an int and causes a warning or two. It is proably not a dangerous 
> error though.

Could you submit a patch for that?

-aaron