Message132446
| Author |
Rogi |
| Recipients |
Rakeka, Rogi, eric.frederich, georg.brandl, mhammond, santoso.wijaya, vstinner |
| Date |
2011年03月29日.00:10:07 |
| SpamBayes Score |
1.4234397e-06 |
| Marked as misclassified |
No |
| Message-id |
<1301357408.06.0.608626170863.issue6498@psf.upfronthosting.co.za> |
| In-reply-to |
| Content |
@mhammond
Maybe its just me but it seems to be a really bad idea to let
those functions terminate your process as they wish. Teh programmer
should be allowed to control teh flow of his application.
However, I am not teh only one affected by this. I recall seeing
blender crash when I typed raise(SystemExit) on its interactive console.
Teh new blender beta worked around this by UNDEFINING SystemExit, which,
to me, seems a UGLY HACK.
If backwards compatibility is teh problem, things could still be
hacked to maintain it and achieve teh desired behaviour during
transition. For example, one could set a macro GOOD_BEHAVING_PYTHON and
obtain teh desired behaviour, without breaking old stuff. It's ugly but
it works.
Teh new behaviour could be something liek that:
...
#define GOOD_BEHAVING_PYTHON 1
#include <Python.h>
...
PyRun_SimpleString(...);
/* allow programmer to do what he wants */
/* if he wants to, he can do teh following */
PyErr_Print();
PyErr_Handle();
...
this way, Py_Main() could be implemented in a way it would return to its
caller instead of terminating teh process.
Adn about my keyboadr, its not juts teh 'h' adn 'e' keys.
Cheers. |
|