Guido> I'm sure the 'struct' module is implicated only because it happens to
Guido> be the first module that setup.py tries to build.
Guido> This points to a problem with the dynamic linker.
Yes, it does. Indeed, if I look in Python/dynload_shlib.c, near
the end, I see code like this:
if (Py_VerboseFlag)
printf("dlopen(\"%s\", %x);\n", pathname, dlopenflags);
handle = dlopen(pathname, dlopenflags);
if (handle == NULL) {
PyErr_SetString(PyExc_ImportError, dlerror());
return NULL;
}
If I insert, immediately after the call to dlopen, the following:
if (Py_VerboseFlag)
printf("after dlopen(\"%s\", %x);\n", pathname, dlopenflags);
fflush(stdout);
and then run "./python -v" and try to import struct, it does not print
the second set of output:
$ ./python -v
<lots of output>
[GCC 3.1.1] on sunos5
Type "help", "copyright", "credits" or "license" for more information.
import readline # builtin
>>> import struct
dlopen("/export/spurr1/homes1/ark/test-python/Python-2.2.1/build/lib.solaris-2.7-sun4u-2.2/struct.so", 2);
Segmentation Fault - core dumped
This behavior strongly suggests to me that it is crashing in dlopen.
However, when I write a little C program that just calls dlopen with the
file in question:
#include <stdio.h>
void *dlopen(const char *, int);
main()
{
void *handle = dlopen("/export/spurr1/homes1/ark/test-python/Python-2.2.1/build/lib.solaris-2.7-sun4u-2.2/struct.so", 2);
printf ("Handle = %x\n", handle);
}
it quietly succeds and prints "Handle = 0"
At this point I am way out of my depth.
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4