Etienne Robillard <e...@gthcfoundation.org> added the comment: On 03/01/11 09:45 AM, R. David Murray wrote: > R. David Murray <rdmur...@bitdance.com> added the comment: > > Etienne: since this is about solving a 3.x specific problem, it will not get > backported. Issue 1573931 looks unrelated to me at a quick glance. FYI, you > will find that you *do* have detach in 2.7 if you open a file using the io > subsystem (import io). Of course, that isn't used for the std files in 2.7. > > Glen: the new IO subsystem is a complete C layer on top of only the most > basic of the C runtime stuff. It does handle cross platform issues. Given > that, and given that the input to CGI *should* be bytes, I think letting an > error raise if the stream is text and detatch isn't available is fine, though > we might find we want to catch it to improve the error message with extra > context. > > Pierre: yes, that diff is what I was looking for. I hope to have time to > look it over later today. > > ---------- > > _______________________________________ > Python tracker <rep...@bugs.python.org> > <http://bugs.python.org/issue4953> > _______________________________________ >
Thanks for theses precisions, David. So will cgi.FieldStorage still be usable in 3.x using 2.5 semantics ? implementing the size argument in the FieldStorage class would surely be a good fix for WSGI middlewares. Either ways (using the new io subsystem) or monkey-patching cgi.FieldStorage so it accepts the size argument could probably helps to resolve memory-usage issues with things like file uploads! Regards ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue4953> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
