On Tue, Jul 19, 2005, Dan Fandrich <dan_at_coneharvesters.com> said:

> On Tue, Jul 19, 2005 at 09:05:41AM -0400, Robert P. J. Day wrote:
> [...]
>> and the subsequent "make" worked. now, i would have interpreted the
>> above to mean that, say, i had no krb4 support, so i should certainly
>> be able to add "--without-krb4" to my configure step, right? (yes,
>> it's redundant, but i'm just being pedantic.)
>>
>> if i do this, the configure step completes but, partway through, you
>> can read:
>>
>> ...
>> checking if argv can be written to... yes
>> checking if Kerberos4 support is requested... yes
>> checking where to look for Kerberos4... libs in no/lib, headers in no/include
>> ...
>>
>> huh? i specifically said i *didn't* want krb4 support, so what does
>> this mean? in addition, even though the configuration completes, the
>> make eventually fails with:
>
> Looks like a bug in the configure script. The situation when a --without-*
> flag is given is handled differently internally from a situation where
> no flag is given at all. As you might guess, the --without-* flag sets an
> option to the magic string "no" whereas no option at all sets it to the ""
> empty string. It seems the latter is handled but the former is not for
> kerberos4 and gnutls.
>
>> can anyone explain what's going on here, and why the configuration
>> process is so confusing and inconsistent?
>
> While this should be fixed in the curl configure script, the underlying
> inconsistency is autoconf's fault. I see no reason that a --without flag
> (or --with, depending on what the default should be) should be treated any
> differently than when the flag isn't given at all.

IIRC, it is possible for "" to be considered 'yes' if the program wishes.
in this case --without and "" are not the same thing. or have i
overestimated the capabilites of autoconf?

allan

>
>>>> Dan
> --
> http://www.MoveAnnouncer.com The web change of address service
> Let webmasters know that your web site has moved
>

--