Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Message-Id: <5.1.0.14.2.20021229175029.03942b90@pop3.rowe-clan.net>
Date: Sun, 29 Dec 2002 17:58:31 -0600
To: <rbb@apache.org>
From: "William A. Rowe, Jr." <wrowe@apache.org>
Subject: Re: cvs commit: apr/include apr_errno.h
Cc: <dev@apr.apache.org>
In-Reply-To: <Pine.LNX.4.33L2.0212291548240.23455-100000@vangogh.rkbloom
 .net>
References: <5.1.0.14.2.20021229172147.033f16d8@pop3.rowe-clan.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

At 05:50 PM 12/29/2002, rbb@apache.org wrote:

>On Sun, 29 Dec 2002, William A. Rowe, Jr. wrote:
>
>> This patch comes up just a bit short.
>>
>> Because a user *may* choose to use the explicit value APR_ESYMNOTFOUND,
>> the test APR_STATUS_IS_ESYMNOTFOUND must always include it, as well
>> as any platform-specific values.
>
>What user?  The only thing that ever sets a value to ESYMNOTFOUND is APR
>itself.  On Windows, we don't use that value, so we don't test for it.

You've never used the construct;

    if (xxxcondition) {
        rv = APR_EFOO;
    }
    else {
        rv = apr_foo_fn(somefn);
    }

to express the fact that you are skipping a call?  If APR_EFOO is defined,
then APR_STATUS_IS_EFOO(APR_EFOO) must resolve to true.

Of course you could have simply #defined APR_EFOO as PLATFORM_RES
and therefore satisfied the requirement above and your desire to fix the test.