IE10 download failures from US Patent and Trademark Office

Problem:

Downloads from the USPTO.GOV site, specifically the "PAIR" patent application retrieval site, are failed by IE10.

Steps to repro:

• Example Setup: Win8 RTM (64 bit), desktop IE10, (either proxy or Server 2012 NAT used), NetMon for traces
• Start at http://portal.uspto.gov/external/portal/pair
• Select “Public PAIR”
• go through CAPTCHA
• Lookup application # 11/536,573 (or recent patent, or ...)
• Select the tab "Image File Wrapper"
• Select about 10-20% of the files using the Checkboxes on the right column under the "PDF" (this ensures % downloaded will go above zero, showing at least some progress)
• Click the blue "PDF" icon to initiate the download
• OPTION 1: Choose to open the file directly
  1. Result: it will download the selected portions, then report "[name].pdf couldn't be downloaded."
• OPTION 2: Choose to save the file to desktop
• Result: it will save a portion to the directory specified, such as "[name].pdf.[random].partial"

1. NOTE:  if nothing at all downloads, choose another application number, via the “Select New Case” tab, and continue from step #4)
2. The file will again report as a failed download
3. IMPORTANT: if the file is renamed, the file will open properly – e.g., copy 11536573.pdf.agavx0o.partial 11536573.pdf, and open it

Expected results:

Downloads succeed, even if told a size larger than the actual data size provided, so long as the download finishes "cleanly".

Possible cause/related:

When requesting a file, it appears the site calculates a fixed maximum size per page to be sent, because it generates the PDF on-the-fly from TIFF images.  The site sends the generated PDF as the downloaded file.  After the last of the file is downloaded, the site terminates the TCP connection for the download (TCP.Flags.FIN=1), effectively truncating the file size.  IE10 treats this as a failed download.

Question:

Is there any way to workaround this problem when using IE10?

Thanks...

Note: NetMon captures are available to Microsoft upon request.  Capture shows the server as being IBM_HTTP_SERVER.  PDF headers show creator of the PDF as "USPTO PDF Builder".  Again, the resulting PDF is valid, if only IE10's download manager could treat it as such....

10 years Windows storage drivers | Microsoft Patent Lawyer | LCA - Patents | Microsoft (not giving legal advice)