Re: [build] status on Mac OS X 10.6.2

[Alexander Hansen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 1/10/10 5:58 PM, Ben Abbott wrote:
> On Jan 10, 2010, at 3:30 PM, Alexander Hansen wrote:
> 
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 1/9/10 9:49 PM, Ben Abbott wrote:
>>> On Jan 9, 2010, at 12:21 AM, Michael D Godfrey wrote:
>>>
>>>> The attached file documents the steps that
>>>> led to a reasonably functional install on my
>>>> MacBook Pro running OS X 19.6.2.
>>>>
>>>> The remaining work of installing additional packages
>>>> should be pretty straightforward.
>>>>
>>>> Aside from boring things like sed and flex, the somewhat
>>>> surprising problem was that none of the available
>>>> Fortran compilers worked reliably with -O2 (actually
>>>> I changed all of the compiler options from -O2 to -O0
>>>> so the C and C++ compilers could also be at fault).
>>>>
>>>> For now I plan to leave the optimization set to -O0.
>>>> When things are more settled will be good to try -O2
>>>> again.
>>>>
>>>> Michael
>>>>
>>>> <Octave_install_8_Jan_2010.txt>
>>>
>>> As a follow up to Michael's experience, I've documented my experience using 
>>> the Fink package manager to satisfy the dependencies.
>>>
>>> I've changed the subject, as my approach does not presently result in 
>>> installing octave.
>>>
>>> Ben
>>>
>>> p..s I've cc'd Alexander Hansen, who is Fink's package maintainer for 
>>> Octave. Alex, I thought this would be of interest to you.
>>>
>> The bit at the end saying "Eventually, a Fink package will be available.
>> However, that will require a new gcc-4.2 package that includes
>> gfortran." isn't correct.  Our existing octave builds using our gcc44
>> compiler rather than Xcode.
>>
>> What's holding up official Fink updates is a lack of my time (much of
>> which is occupied using octave for analysis and I don't want to break my
>> scripts in the process--octcdf didn't get along with Octave-3.2 in my
>> testing).
>> - -- 
>> Alexander Hansen
>> Fink User Liaison
> 
> Hi Alex,
> 
> With Octave's current sources, I gave up trying gcc 4.4 to build. The problem 
> I encountered was a "bus-error" when mixing gcc 4.4 with gcc 4.2. We were 
> able to isolate the problem to linking to the OpenGL and Carbon frameworks. 
> Meaning that if we did not link to those, there was no bus-error.
> 
> I suspect the same problem would be encountered if FLTK were build with gcc 
> 4.4. However, at the moment, I noticed there are compiler errors if gcc 4.4 
> is used to build FLTK. So I'm just speculating the a "bus-error" would occur 
> if FLTK is built with gcc 4.4.
> 
> In any event, you obviously have much more experience with the details of 
> building apps for MacOS X. I am optimistic you'll find a more desirable 
> solution that abandoning gcc 4.4 and backing up to gcc 4.2.
> 
> Ben
> 

Right, considering that any FSF gcc older than 4.4 appears not to work
properly for 64-bit stuff (Apple's is another matter, of course, but no
Fortran there).


- -- 
Alexander Hansen
Fink User Liaison
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAktKYPsACgkQB8UpO3rKjQ/mcgCfU1aoNe9740YMBaXeyoWX+w+A
gqcAnjgAm/dmVUMEl4jyT5vThXXc+Wfh
=tflx
-----END PGP SIGNATURE-----