For what it's worth, I agree with Lachele here. I agree with Jason that most of the people who have a legitimate reason to have $AMBERHOME set to some other value are developers. But, clearly, even we can be flummoxed by the symptoms of a bad $AMBERHOME. And what of the "normal user" who commits a typo when setting $AMBERHOME?
Having said that, I don't think configure should force the user to set the "correct" $AMBERHOME. For which reason, I like Lachele's idea of an "are you sure" question.
Ben
--
For greater security, I support S/MIME encryption.
Sent from my iPhone
On 8/04/2011, at 7:26 AM, "B. Lachele Foley" <lfoley.uga.edu> wrote:
> I typically leave amberhome unset until needed. If not set, the last time I did that, the compile fails after a long time with an unhelpful error message. I'd rather it die straight away. I would appreciate a check for having it set incorrectly that says "are you sure?" if it seems wrong.
>
> I'll put that on my list. :-)
>
>
>
> Sent from my iPhone
>
> On Apr 8, 2011, at 12:33 AM, Jason Swails <jason.swails.gmail.com> wrote:
>
>> On Thu, Apr 7, 2011 at 9:03 PM, B. Lachele Foley <lfoley.uga.edu> wrote:
>>
>>> I did this, too (but figgered it out). Maybe the compile should fail if
>>> $AMBERHOME is not set? That might do a lot of good in general -- alert
>>> users to the importance of the setting.
>>>
>>
>> It's easy enough to check that AMBERHOME is set -- but that it's set
>> correctly? You could drop a check in configure to make sure that it's not
>> set *incorrectly* and force you to either set it properly or unset it, but
>> what's the benefit here? How many users do we think will actually have
>> cause to set it incorrectly (i.e. at a different amber) besides developers?
>>
>> All the best,
>> Jason
>>
>>
>>> Sent from my iPhone
>>>
>>> On Apr 7, 2011, at 11:54 PM, Thomas Cheatham <tec3.utah.edu> wrote:
>>>
>>>>
>>>> Re: "my bad" $AMBERHOME not set appropriately...
>>>>
>>>> (bad news: experienced user == trusting idiot / not so experienced)
>>>> (good news: we now know what happens if you do not set $AMBERHOME)
>>>>
>>>> I apologize for the hassle and frustration especially as release is
>>>> pending. What I did was simply extract the tarball, configure intel and
>>>> make install. Nothing else. I uncovered the python issue and then
>>>> started to blindly dig on multiple machines (like kraken, exposing other
>>>> issues that we still should resolve).
>>>>
>>>> I think, however, that Jason nailed it. I wasn't setting $AMBERHOME and
>>>> had assumed that it was automagically set. As this had been set
>>>> automatically previously, I assumed it was set in the current install
>>>> scripts and I was incorrect in this assumption. [It may be beneficial
>>>> behavior however to reset AMBERHOME automatically however!].
>>>>
>>>> The only upside is that perhaps we now know what errors people will see
>>> if
>>>> they do not set AMBERHOME correctly; perhaps this has not been a total
>>>> waste of all of our time...
>>>>
>>>> I went back and did it all over again, this time with $AMBERHOME set
>>>> appropriately and after the wait for recompile, it all worked fine.
>>>>
>>>> --tom
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>>
>>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Apr 08 2011 - 05:00:02 PDT