A non-fink version would be easier to distribute over many identical
Say some one wants to install analysis on 12 machines in a lab;
Currently they would have to install fink on each machine, then all
the development libraries, then compile analysis. This would take a
long time and involve much repetitive work on many machines.
As Apple systems are quite homogenous in that their kernel /
development tools are identical (when fink is *not* installed), then a
distribution without fink would allow you to compile on one machine,
and then only distribute that binary to all machines.
On 27 May 2008, at 12:53, Murali Vadivelu wrote:
> Dear Wayne,
> Out of curiosity, I wonder why some people want to avoid Fink.
> Many thanks.
> Best regards,
> On 20 May 2008, at 23:53, Wayne Boucher wrote:
>> Ah, I'm glad that worked. I had thought you wanted an Aqua version
>> I'd like to provide but haven't figured out how to yet). For non-
>> compiling things as you say should indeed work. In fact I think
>> (if I
>> remember correctly) I did this for someone in Cambridge last week
>> who also
>> didn't want to use Fink. The one thing that needed doing that was
>> was to edit the Modules/Setup file to make sure that Tkinter is
>> being (a)
>> compiled and (b) picking up the correct Tcl/Tk.
>> (In my spare moments I'm slowly figuring out how to package this
>> all up on
>> the Mac so it will hopefully just be a few click procedure in
>> Regards, Wayne
>> On Tue, 20 May 2008, Ben Eisenbraun wrote:
>>> Hi Wayne,
>>> (Backstory: I'm trying to build a standalone version of CCPNMR for
>>> OS X without using Fink.)
>>> On Wed, Apr 30, 2008 at 10:29:11AM +0100, Wayne Boucher wrote:
>>>> PS: It's possible this is down to not choosing the correct Mac
>>>> compile-time options, but there could also be something deeper
>>>> I'm not sure.
>>> Yes, that was it.
>>> If you want to build a standalone version for CCPNMR on OS X, you
>>> to build non-framework versions of Tcl/Tk, and then a version of
>>> Python that links against those Tcl/Tk libraries and _not_ against
>>> Aquafied Tcl/Tk libs in /usr/lib. The only painful bit is that
>>> you have
>>> to patch Python's setup.py in order to do that.
>>> With those built, you can hack up installCode.py to point to the
>>> locations for your non-system Tcl/Tk and python, and it builds and
>>> Thanks for your guidance.
>>> Ben Eisenbraun
>>> Structural Biology Grid Harvard Medical
>>> http://sbgrid.org http://hms.harvard.edu
[log in to unmask]