Author loewis
Recipients ajaksu2, amaury.forgeotdarc, bob.ippolito, georg.brandl, loewis, pitrou, rhettinger
Date 2009-02-28.09:28:12
SpamBayes Score 4.82475e-06
Marked as misclassified No
Message-id <0016e6434492409f1f0463f734b8@google.com>
In-reply-to
Content
The one thing that IMO needs to be decided before this can be accept is
the version compatibility: what Python versions must this code stay
compatible with? That decision then needs to be implemented.

Apart from this (and the additional minor comments below), the patch
looks fine.

http://codereview.appspot.com/20095/diff/1/13
File Lib/json/decoder.py (right):

http://codereview.appspot.com/20095/diff/1/13#newcode21
Line 21: nan, inf = struct.unpack('dd', _BYTES)
I think this can be simplified as

   nan, inf = struct.unpack('>dd', _BYTES)

http://codereview.appspot.com/20095/diff/1/12
File Lib/json/encoder.py (right):

http://codereview.appspot.com/20095/diff/1/12#newcode31
Line 31: INFINITY = float('1e66666')
Why not decoder.PosInf?

http://codereview.appspot.com/20095/diff/1/14
File Modules/_json.c (right):

http://codereview.appspot.com/20095/diff/1/14#newcode3
Line 3: #if PY_VERSION_HEX < 0x02060000 && !defined(Py_TYPE)
Is Python before 2.6 even supported anymore? ISTM that the usage of
.format on strings outrules Python2.5 and earlier.

http://codereview.appspot.com/20095
History
Date User Action Args
2009-02-28 09:28:16loewissetrecipients: + loewis, georg.brandl, rhettinger, bob.ippolito, amaury.forgeotdarc, ajaksu2
2009-02-28 09:28:14loewislinkissue4136 messages
2009-02-28 09:28:12loewiscreate