[Python-Dev] Re: Bug [ 959379 ] Implicit close() should check for errors

[Peter Astrand]

On Tue, 26 Oct 2004, "Martin v. Löwis" wrote:

> > Yes. Thanks. Here's the new patch. OK to commit?
>
> No. I feel that perror is inappropriate; PySys_WriteStderr
> might be slightly better.

Here's a patch with PySys_WriteStderr:

diff -u -r2.192 fileobject.c
--- Objects/fileobject.c        11 Jun 2004 04:49:03 -0000      2.192
+++ Objects/fileobject.c        26 Oct 2004 18:44:19 -0000
@@ -300,12 +300,15 @@
 static void
 file_dealloc(PyFileObject *f)
 {
+       int sts = 0;
        if (f->weakreflist != NULL)
                PyObject_ClearWeakRefs((PyObject *) f);
        if (f->f_fp != NULL && f->f_close != NULL) {
                Py_BEGIN_ALLOW_THREADS
-               (*f->f_close)(f->f_fp);
+               sts = (*f->f_close)(f->f_fp);
                Py_END_ALLOW_THREADS
+               if (sts == EOF)
+                       PySys_WriteStderr("close failed: %s\n", strerror(errno));
        }
        PyMem_Free(f->f_setbuf);
        Py_XDECREF(f->f_name);


>However, it might be that even
> raising an exception has some value.

Is it really possible to raise an exception in response to something
triggered by the GC?

/Peter Åstrand <astrand at lysator.liu.se>