В Mon, 19 Sep 2016 15:32:56 +0200
Dima Tisnek <dimaqq_at_gmail.com> пишет:

> Indeed, it looks like `Curl` is designed to be a throw-away object.
>
> If it were intended for reuse, I'd expect to see all the default
> settings in own method, which is called from {init, get, post}.
>
> On 19 September 2016 at 15:04, Cepгей А via curl-and-python
> <curl-and-python_at_cool.haxx.se> wrote:
> > Hello,
> > There is a bug in pycurl`s curl module. When using
> > curl.Curl requests subsequently, curl.Curl.__request returns the
> > current request` body, inculing the previous bodies. The cause is
> > about wrong handling of curl.Curl`s internal StringIO buffer. It
> > can be fixed by adding these two lines to curl.Curl`s __request
> > function:
> >
> > self.payload_io.seek(0)
> > self.payload_io.truncate()
> >
> > I`m running Python 2.7 ,PycURL/7.43.0 libcurl/7.50.1 on Debian Sid.
> >
> > _______________________________________________
> > https://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python

They didn`t tell that it is a throwaway thing. I think is`s just a
quick and simple interface to the pycurl library. Which can and should
be reused. Currently, the fact that __request does not clear the buffer
prevents it from being reused. However, I don`t think all the defaults
should be called from requests. Just one time, during the
instantiation, because you`ll have to redefine the options in some
cases.

As I recall, the whole thing (multiple requests) was working alright a
while ago. Did something change? If not, would anyone care to patch it?

_______________________________________________
https://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-python
Received on 2016-09-20