Hello,
We upgraded recently JMeter in a branch from HttpClient 4.2.6 to 4.5 and
are facing what seems to be a regression.
It seems that when we don't set a charset in StringBody , MultipartEntity,
FormBodyPart, no charset is written in ContentType generation while it was
the case at least in 4.2.6, I think issue was introduced in 4.3.
Note also that javadocs of these components say:
@param charset the character set to use, may be {@code null}, in which case
{@link MIME#DEFAULT_CHARSET} - i.e. US-ASCII - is used.
But it's not the case since 4.3
Reading the sepcs, I am not sure that when charset is US-ASCII , it should
be written but it was the case in 4.2.6.
I created https://issues.apache.org/jira/browse/HTTPCLIENT-1665
As a test case, you can have a look at JMeter test cases:
org.apache.jmeter.protocol.http.sampler.TestHTTPSamplersAgainstHttpMirrorServer#testPostRequest_FormMultipart3
org.apache.jmeter.protocol.http.sampler.TestHTTPSamplersAgainstHttpMirrorServer#testPostRequest_FileUpload3
The issue appears when no contentEncoding is set.
Thanks for help.
Regards
Philippe M.
