Reproduced a test case for this that fails on Chrome but passes on Safari. Reported in the Chromium tracker as an extension of a very similar bug:
http://code.google.com/p/chromium/issues/detail?id=79556 On Thu, Jun 23, 2011 at 9:54 AM, Paul Davis <[email protected]> wrote: > I'm not entirely certain if the bug is in Chrome or how we expect > Chrome to react. I know there are browser differences when running > against localhost but I'm not entirely certain what they are and why. > I'll write a small test case in Python to try and reproduce and if so > will call it a Chrome bug and kick it to their issue tracker. > > On Thu, Jun 23, 2011 at 5:02 AM, Randall Leeds <[email protected]> > wrote: >> Does this affect anything serious or just a contrived example? Is it a >> zero-length attachment or something, in the test? >> >> File it on the chromium issues and call a vote for 1.0.3 :) >> On Jun 22, 2011 5:58 PM, "Paul Davis" <[email protected]> wrote: >>> While preparing the 1.0.3 release I stumbled across an awesome bug in >>> the attachments.js test. Apparently Chrome will set the Content-Type >>> to null if the Content-Length is 0 and the response is served from >>> 127.0.0.1. I can reproduce this easily by just changing between >>> 127.0.0.1 and my local external IP 192.168.10.x in the URL bar and run >>> the test on both. Fails on 127.0.0.1 and not on the external IP. This >>> appears to be consistent on all branches. >>> >>> I have no idea what the best fix is here. Suggestions welcome. >> >
