Hi Sven, > -1 > > #asByteArray does an implicit text to bytes conversion (granted, a stupid > one). > This will (and had conflicted in the past) with the text conversion done > automatically by the HTTP client. > The response simply has to be binary because the data is binary.
Yes, I understand. The important thing is that at the moment is that a misconfigured server results in something that gives the appearance of a Pharo bug. >> Failing that, the condition should probably be explicitly checked for >> and a more useful error message given, as the cause is far enough away >> from the symptom that it takes a bit of digging to find out what went >> wrong. > > +1 > > Yes, the error should be reported better. As it is today is not acceptable. > > Normally, the ZnClient>>#accept: option (in combination with #systemPolicy) > would do the trick. > However, while a ZnMimeType text wildcard (text/*) is useful to require > textual responses, no such general wildcard exists for (all) binary > responses. We could try with (application/*) and hope that enough MC servers > comply (I think most will use application/octet-stream or > application/x-monticello anyway). > > I could try to make a 2.0 patch and see what happens. Excellent! Regards, Stuart