Pickling a httplib.HTTPMessage fails <1% of the time for URLs with spaces

Asked Jul 11 '12 at 03:19

Active Jul 11 '12 at 03:19

Viewed 291 times

In Python 2.7.2 pickling a httplib.HTTPMessage works literally more than 99% of the times. But sometimes on rare occasions, it fails:

dumps(info)

TypeError: can't pickle StringO objects


dumps(info, 2)

PicklingError: Can't pickle <type 'cStringIO.StringO'>: attribute lookup cStringIO.StringO failed

How can I find out what member of the httplib.HTTPMessage class (which info is an instance of) has the problem?

I got info from calling open() on an object returned from urllib2.build_opener. All URLs which have failed so far have spaces in them.

asked Jul 11 '12 at 03:19

Petter

37,121
7
47
62

So you're saying that if you take a single object and pickle it 100 times in a row, 1 of those times it will fail to pickle? – Gabe Jul 11 '12 at 03:22
Gabe: No, I am pickling different objects obtained by downloading different URLs. Most of the time it works. – Petter Jul 11 '12 at 03:27
So then is it certain URLs that don't work while others always do? – Gabe Jul 11 '12 at 11:46
Can you post a URL that fails? – phihag Jan 10 '13 at 18:05

Pickling a httplib.HTTPMessage fails <1% of the time for URLs with spaces

0 Answers0