I was trying to understand why in this archive http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/browser messages from Mattmann, Chris A (388J) do not load, such as the second. Using firefox, if I open link in new tab, I get: http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/ajax/%3CC8F432CA.238DB%Chris.A.Mattmann@jpl.nasa.gov%3E which has the obvious perecent escape problem of not having escaped the % in the message id. Fix this and http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/ajax/%3CC8F432CA.238DB%25Chris.A.Mattmann@jpl.nasa.gov%3E returns appropriate XML content c.f. http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/%3CC8F432CA.238DB%25Chris.A.Mattmann@jpl.nasa.gov%3E The headers say: Server: Apache/2.3.8 (Unix) mod_ssl/2.3.8 OpenSSL/1.0.0a
It seems more complicated: examine the source for http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/%3CC8F432CA.238DB%25Chris.A.Mattmann@jpl.nasa.gov%3E at the end is: <li>Unnamed multipart/alternative (inline, None, 0 bytes)</li> <ul> <li><a rel="nofollow" href="/mod_mbox/incubator-general/201011.mbox/raw/<C8F432CA.238DB%Chris.A.Mattmann@jpl.nasa.gov>/1">Unnamed text/plain</a> (inline, Quoted Printable, 5490 bytes)</li> </ul> </ul> </td> </tr> <tr class="raw"> <td class="left"></td> <td class="right"><a href="/mod_mbox/incubator-general/201011.mbox/raw/%3cC8F432CA.238DB%25Chris.A.Mattmann@jpl.nasa.gov%3e" rel="nofollow">View raw message</a></td> </tr> </tbody> </table> </body> </html> Notice the first link is bad (and 400s), the second is good (and 200s). So the subtleties in %-escaping % seem to have defeated the programmer here.
This should be fixed in r1187588, but I am waiting with closing the bug until that is installed on mail-archives.apache.org
This is fixed now on mail-archives.apache.org
*** Bug 52039 has been marked as a duplicate of this bug. ***
The Chris Mattmann mails no longer open in Firefox or Chrome in browse mode. Has the fix been lost?
Still fails in Chrome 15.0.874.106 m But has started working in Firefox 8.0. This appears to be because that fixes the bad characters in the URL. So it looks like the fix is still lost.
According to https://issues.apache.org/jira/browse/INFRA-4155, mail-archives.a.o - eos - runs r1188408. However the Mattman mail still generates an incorrect link. http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/browser generates: http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/ajax/%3CC8F432CA.238DB%25Chris.A.Mattmann%40jpl.nasa.gov%3E whereas http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/thread?0 generates http://mail-archives.apache.org/mod_mbox/incubator-general/201011.mbox/%3cC8F432CA.238DB%25Chris.A.Mattmann@jpl.nasa.gov%3e
Fixed with the merge of the convert-charset branch, r1326076 Now live on mail-archives.apache.org
On mail-archives.apache.org, we now see a message that requires double-escaping: http://mail-archives.apache.org/mod_mbox/oodt-dev/201304.mbox/%3CCD80E8CA.CEA8%2525Daniel.J.Crichton@jpl.nasa.gov%3E At a guess, that is because the original message-id contains a percent sign followed by two hex digits. (In the original example, the first two letters of "Chris" are not both hex digits.) See: https://issues.apache.org/jira/browse/INFRA-6244
Remaining case fixed with r1484915. The msgId was percent decoded twice in the module, so that a percent sign in the msgID resulted in trouble. Since the msgID was derived from path_info which was already percent decoded, there was no reason to percent decode the msgID itself again. All links including a msgID should have already been single encoded before this commit so work well with single decoding. Code is running on US instance of mail-archives.apache.org. Will update the EU instance during the next few hours as well.