ports/168214: Heimdal 1.5.2 problem

[Joerg Pulz]

On 25.05.2012, at 20:39, Wesley Shields wrote:

> On Fri, May 25, 2012 at 01:20:46PM -0400, Robert Simmons wrote:
>> On Fri, May 25, 2012 at 12:56 PM, Wesley Shields <wxs at freebsd.org> wrote:
>>> On Fri, May 25, 2012 at 12:21:54PM -0400, Robert Simmons wrote:
>>>> On Thu, May 24, 2012 at 8:38 PM, Wesley Shields <wxs at freebsd.org> wrote:
>>>>> On Tue, May 22, 2012 at 06:29:20PM -0400, Robert Simmons wrote:
>>>>>> On Tue, May 22, 2012 at 5:14 PM, Wesley Shields <wxs at freebsd.org> wrote:
>>>>>>> On Tue, May 22, 2012 at 03:08:31PM -0400, Robert Simmons wrote:
>>>>>>>> On Tue, May 22, 2012 at 8:57 AM, Wesley Shields <wxs at freebsd.org> wrote:
>>>>>>>>> As the person who committed this update I will take responsibility for
>>>>>>>>> seeing this through. Would you mind opening a PR with this patch and CC
>>>>>>>>> both myself and the maintainer so it can be properly tracked. I will
>>>>>>>>> work with both of you to make sure it is addressed.
>>>>>>>> 
>>>>>>>> I got some good feedback about the patch. ?I was missing a "\". ?Also,
>>>>>>>> it was noted that I shouldn't make changes to the default settings in
>>>>>>>> this patch since it is meant to correct a problem. ?I removed the
>>>>>>>> change to default.
>>>>>>> 
>>>>>>> I'm not opposed to removing the change to the default, but it does cause
>>>>>>> another problem. See below.
>>>>>>> 
>>>>>>>> Perhaps the different default is not the best solution. ?Maybe there
>>>>>>>> should be a message that at least one backend is needed for the port
>>>>>>>> to function, but none have been selected by default?
>>>>>>> 
>>>>>>> If a backend is required the port should refuse to build if no backend
>>>>>>> is selected. This is pretty easy to do, just check for at least one of
>>>>>>> the backends. I have no idea if multiple backends can be supported so
>>>>>>> you may or may not want to also check for that.
>>>>>> 
>>>>>> I may have been too hasty. ?I've thought of a situation where one
>>>>>> would want to build the port with no backend at all. ?If one wanted to
>>>>>> use the tools in the port to administrate a remote install of Heimdal,
>>>>>> they may want to build it without a backend.
>>>>>> 
>>>>>> My initial thoughts were only for installing the port as a Heimdal
>>>>>> server, and with the --with-berkeley-db=no problem fixed it does not
>>>>>> wrongly find the version of BDB in the base OS. ?With this fix, the
>>>>>> port can function with no backends selected. ?It just won't be able to
>>>>>> function in a server capacity.
>>>>>> 
>>>>>> I am also not an expert in Heimdal, I just installed it from source
>>>>>> via its own instructions and compared that with what the FreeBSD port
>>>>>> was doing. ?I'd wait for the maintainer to make changes to the default
>>>>>> behavior for the above reason.
>>>>> 
>>>>> This all sounds perfectly reasonable to me. :)
>>>>> 
>>>>> If I'm understanding you correctly the patch[1] in ports/168214 is the
>>>>> correct one to commit. The only change I would make is not bumping
>>>>> PORTREVISION since the option is off by default. Sounds like the only
>>>>> thing left to do is wait for maintainer comment on the PR and commit
>>>>> accordingly.
>>>> 
>>>> Sounds good. ?One question: what do you mean by PORTREVISION being off
>>>> by default?
>>> 
>>> There is no need to bump PORTREVISION because the option which you are
>>> changing is off by default so there's no need to force a rebuild of it
>>> on the package cluster since your changes are going to have no effect
>>> there.
>>> 
>>> For those that have the option to on, it hasn't built properly for them
>>> yet so bumping is going to have no effect either.
>> 
>> I understand what you're saying.  However, my change would actually
>> change the package cluster.  Because those packages were built with
>> "--without-berkeley-db" rather than "--with-berkeley-db=no" the old
>> packages were built with broken BDB support by accident.  By fixing
>> this, the default package is actually going to be different than the
>> one built before this change.  I would recommend bumping PORTREVISION
>> because of this.
> 
> That makes sense. Thanks for the clarification. I will be awaiting
> maintainer approval or timeout then.

Hi,

please go ahead and commit and close ports/168214 using the last version of the patch and please bump PORTREVISION.
Sorry for this, i didn't noticed the configure arg change for building without berkeley-db.

Kind regards and thanks
Joerg