Created on 2012-07-26 18:35 by theller, last changed 2012-08-03 02:58 by meador.inge.
| Files | ||||
|---|---|---|---|---|
| File name | Uploaded | Description | Edit | |
| showbug.py | theller, 2012-07-26 18:35 | demonstrates the bitfield bug | ||
| bitfields.patch | theller, 2012-07-26 18:38 | review | ||
| Messages (3) | |||
|---|---|---|---|
| msg166496 - (view) | Author: Thomas Heller (theller) *  |
Date: 2012-07-26 18:35 | |
When a ctypes Structure uses a subclass of c_int, c_short, c_byte in the _fields_ list to define bitfields, the results are wrong. The attached script showbug.py demonstrates this behaviour. |
|||
| msg166497 - (view) | Author: Thomas Heller (theller) * |
Date: 2012-07-26 18:38 | |
The attached patch for branch default fixes this issue. It also contains a unittest. |
|||
| msg167285 - (view) | Author: Meador Inge (meador.inge) * |
Date: 2012-08-03 02:58 | |
OK, I finally had some time to look at this. The attached patch does fix the issue. However, it has a nasty side effect in that the encoded bitfield size stays with the field member object:
>>> class SubByte(c_ubyte): pass
...
>>> class A(Structure):
... _fields_ = [("x", SubByte, 5)]
...
>>> sizeof(A().x)
327680
Before the patch things look like:
>>> class SubByte(c_ubyte): pass
...
>>> class A(Structure):
... _fields_ = [("x", SubByte, 5)]
...
>>> sizeof(A().x)
1
Leaking the internal bitfield size encoding seems like a deal breaker to me.
|
|||
| History | |||
|---|---|---|---|
| Date | User | Action | Args |
| 2012-08-03 02:58:45 | meador.inge | set | type: behavior messages: + msg167285 |
| 2012-07-26 21:16:51 | eric.araujo | set | nosy:
+ amaury.forgeotdarc, belopolsky, meador.inge stage: patch review versions: + Python 3.2, - Python 3.4 |
| 2012-07-26 19:10:08 | Arfrever | set | nosy:
+ Arfrever |
| 2012-07-26 18:38:47 | theller | set | files:
+ bitfields.patch messages: + msg166497 |
| 2012-07-26 18:35:05 | theller | create | |