On Mon, Oct 4, 2010 at 11:11 AM, Greg Parker wrote:
> On Oct 2, 2010, at 6:54 PM, Kyle Sluder wrote:
>
> Maybe weak linking can help here. If you weak link the Python library and
> call dlopen early on in your program's lifetime with the appropriate path, I
> think dyld will populate all the weak
On Oct 2, 2010, at 6:54 PM, Kyle Sluder wrote:
> Maybe weak linking can help here. If you weak link the Python library and
> call dlopen early on in your program's lifetime with the appropriate path, I
> think dyld will populate all the weak linked functions it finds.
No, weak linking doesn't do
On 3. Oct 2010, at 3:54, Kyle Sluder wrote:
> On Oct 2, 2010, at 3:10 AM, Kurt Sutter wrote:
>
>> Sorry to be obtuse, but it may be that I am overlooking something.
>> Presently, we have calls such as
>>
>> n = PyNumber_Float(obj)
>> s = PyString_AsString(obj)
>>
>> I understand that I would
On Oct 2, 2010, at 3:10 AM, Kurt Sutter wrote:
> Sorry to be obtuse, but it may be that I am overlooking something. Presently,
> we have calls such as
>
> n = PyNumber_Float(obj)
> s = PyString_AsString(obj)
>
> I understand that I would need to generate #defines for all those names, such
> a
On Oct 2, 2010, at 7:20 AM, Ken Thomases wrote:
>> Sorry to be obtuse, but it may be that I am overlooking something.
>> Presently, we have calls such as
>>
>> n = PyNumber_Float(obj)
>> s = PyString_AsString(obj)
>>
>> I understand that I would need to generate #defines for all those names,
>>
On Oct 2, 2010, at 5:10 AM, Kurt Sutter wrote:
> Sorry to be obtuse, but it may be that I am overlooking something. Presently,
> we have calls such as
>
> n = PyNumber_Float(obj)
> s = PyString_AsString(obj)
>
> I understand that I would need to generate #defines for all those names, such
> as
On 2 Oct 2010, at 11:07, Ken Thomases wrote:
> On Oct 2, 2010, at 3:48 AM, Kurt Sutter wrote:
>
>> Yes, that is an option I have considered. But, if I understand this
>> correctly, it would mean resolving some hundred symbols through dlsym and
>> providing a glue function for each of them. Rig
On Oct 2, 2010, at 3:48 AM, Kurt Sutter wrote:
> Yes, that is an option I have considered. But, if I understand this
> correctly, it would mean resolving some hundred symbols through dlsym and
> providing a glue function for each of them. Right?
I don't see the need for a glue function. You do
Ken
Yes, that is an option I have considered. But, if I understand this correctly,
it would mean resolving some hundred symbols through dlsym and providing a glue
function for each of them. Right?
(One other option I have just discovered is to place a symlink file in ~/lib,
which seems to be i
On Oct 1, 2010, at 11:49 PM, Kurt Sutter wrote:
> In our application, the user should be able to change the version of a
> library that the application is linked to.
Don't link to the library/framework. Dynamically load it (dlopen, dlsym,
etc.), instead. Granted, this is a pain in the neck co
In our application, the user should be able to change the version of a library
that the application is linked to. Specifically, our application links to the
Python library. However, some users want to use the Python at
/System/Library/Frameworks/Python.framework/Versions/2.6, while others may wa
11 matches
Mail list logo
