Thank you Geert. I think there is some misunderstanding :-) So a few random points:
1/ not sure what you mean by "marking actual file contents with form-data disposition", but yes, the HTML spec and RFC 2388 mandates the HTTP request has the MIME type multipart/form-data, and each entry to be have a `Content-Disposition: form-data` as far as I can see 2/ not sure what `embed` and `attachement` mean (at least, not in the context of a request, for the latter) 3/ these CURL commands are replicating what is sent by a "simple form" which was the starting point for all this 4/ I did replicate it with Postman as well (but it does not allow you to set the Content-Type for one form-data part, as far as I could see, so I could only test with application/octet-stream, and it indeed gets the same result as the corresponding CURL command and what I observe in the form). I am running out of idea :-( Regards, -- Florent Georges http://fgeorges.org/ http://h2oconsulting.be/ On 27 November 2015 at 11:00, Geert Josten wrote: > Yes, multipart/form-data is what you put in HTML, but that doesn’t have to > mean that the actual file contents is marked with form-data disposition. > You also have embed, attachment, etc. Make a simple form, and try to > capture what is sent across the wire exactly, and then see if you can > mimic the exact same with curl or maybe postman.. > > Cheers > > On 11/27/15, 9:34 AM, "[email protected] on behalf > of Florent Georges" <[email protected] on behalf of > [email protected]> wrote: > >>On 27 November 2015 at 06:49, Geert Josten wrote: >> >>> Maybe it is because it is marked as form-data, instead of as attachment. >>> Have you compared with a simple html upload form? >> >>Not sure to understand. Setting enctype="multipart/form-data" is the >>idiomatic, simple way for a HTML form to upload a file, isn't it? >> >>> I¹m sure that worked just fine for me in the past.. >> >>Yup... What bothers me is that the handling of the file content is >>different based on the Content-Type of the corresponding part (read as >>a binary node for application/octet-stream and as a text node for >>text/xml). In that case, I can't make sense of having a text node >>instead of an XML tree for text/xml. >> >>Regards, >> >>-- >>Florent Georges >>http://fgeorges.org/ >>http://h2oconsulting.be/ >>_______________________________________________ >>General mailing list >>[email protected] >>Manage your subscription at: >>http://developer.marklogic.com/mailman/listinfo/general > > _______________________________________________ > General mailing list > [email protected] > Manage your subscription at: > http://developer.marklogic.com/mailman/listinfo/general _______________________________________________ General mailing list [email protected] Manage your subscription at: http://developer.marklogic.com/mailman/listinfo/general
