Adding configure options when installing a port

[Stephen Langer]

On Jun 17, 2010, at 3:17 PM, Scott Webster wrote:

> On Thu, Jun 17, 2010 at 12:08 PM, Ryan Schmidt <ryandesign at macports.org> wrote:
>> 
>> On Jun 17, 2010, at 14:05, Scott Webster wrote:
>>> I'm not entirely sure I agree with disabling by default.  Unless I was
>>> reading this list, I would never have known about this issue, and
>>> would just go with the default.  Disabling openmp would slow down
>>> everything (well, not quite everything I guess) I do with ImageMagick
>>> by a factor of 2.  Same would happen with other ports that can use
>>> openmp if it was disabled there...
>>> 
>>> I guess I'm just saying that there is a trade-off between universal
>>> compatibility and speed here.  I don't have a good handle on how big
>>> of an issue the compatibility thing is.  If it's quite isolated then I
>>> think having the twice (or more) as fast option as default is
>>> reasonable.
>> 
>> Users are expected to read the output of "port variants" to decide if they might want to use any of those variants...
>> 
> 
> Sure... but:
> 
> a) Lots of people probably don't do that, perhaps particularly for
> "common" ports like ImageMagick, or if it is installed as a
> dependency.
> 
> b) Lots of people would not understand that they likely want openmp enabled.
> 
> c) Couldn't this apply almost as well to the people who want openmp
> disabled too?  In Stephen's case his software is not distributed via
> macports.  Is there software in macports that breaks with openmp
> enabled here?  If so, it would be nice for macports to recognize this
> and tell them to use the no-openmp variant, but #126 :(
> 
> Anyway, this is just the way I see it.


I don't know if there are other packages in macports that would break.  I would like to be able to distribute our program from macports, but that's not going to happen soon.  In the meantime, telling our users to install the no-openmp variant of imagemagick would be ok with us.

I'll see if I can get the ImageMagick folks to comment on turning openmp on and off at runtime.

-- Steve


--
-- stephen.langer at nist.gov                    Tel: (301) 975-5423 --
-- http://math.nist.gov/mcsd/Staff/SLanger/   Fax: (301) 975-3553 --
-- NIST, 100 Bureau Drive, Stop 8910, Gaithersburg, Md 20899-8910 --

-- "I don't think this will work.  That's why it's science."      --
--                     Naomi Langer (age 6),  17 Feb 2003         --