For what it's worth, I was also experiencing serious performance problems
with the multipart request handling. I ended up trying out the multipart
handler stuff from "Java Servlet Programming" from Oreilly. I haven't
delved deep into the code so I apologize for not providing any specific
information, but I am able to upload files without any problems using the
examples from Oreilly. Perhaps this is worth looking at for comparisons
(Mark)?
http://www.servlets.com/resources/com.oreilly.servlet/index.html
Scott
-----Original Message-----
From: Stefan Wesner [mailto:[EMAIL PROTECTED]]
Sent: Monday, November 27, 2000 5:30 AM
To: [EMAIL PROTECTED]
Subject: RE: struts.upload, performance and more
I just want to mention that the performance problem exists also using resin
as the servlet engine.
> -----Original Message-----
> From: Schachter, Michael [mailto:[EMAIL PROTECTED]]
> Sent: Monday, November 27, 2000 7:13 AM
> To: '[EMAIL PROTECTED]'
> Subject: RE: struts.upload, performance and more
>
>
> Sorry if this is a duplicate:
>
> > So my question is anyone else has observed performance problems ?
>
> I have, but only with tomcat. The ServletInputStream.readLine() method
> seems to hang on occassion.
>
> >There should be no need for me to bother about
> >MultipartRequestHandler, MultipartIterator and MultipartElement, right
> ?
> >I should just use the FormFile interface and get/setFile methods
> >in my form ?
>
> MultipartRequestHandler is an interface you'd implement to create your
> own way of handling multipart
> requests. MultipartIterator and MultipartElement are helper classes
> that I used to create my own
> implementation of MultipartRequestHandler, they're pretty useful, except
> for this weird performance problem.
>
> >What's the init param in web.xml called that governs where the
> >temporary files are put ?
>
> Either have the javax.servlet.context.tempdir attribute set, or use the
> init param "tempDir"
>
> >It was very easy to get upload of files to work using the
> struts.upload.
> >Now I only hope the performance problem will be fixed..
>
> I'll definately be looking into this problem more.
>
> - Mike Schachter
>
