Hi Joerg,
Good catch! I confirm revision 410112 has a bad log, the correct log
should be related to COCOON-1266.
The diff [1] is just about 1 line:
response.addHeader(Vary, Host);
I dropped the Vary header because the vary header in IE is broken [2].
Hence using it to stop caching is just a hack. See also[3].
Hope this helps.
Best Regards,
Antonio Gallardo.
[1]
http://svn.apache.org/viewvc/cocoon/branches/BRANCH_2_1_X/src/java/org/apache/cocoon/reading/ResourceReader.java?r1=155087r2=410112diff_format=h
[2] http://lists.over.net/pipermail/mod_gzip/2002-December/006826.html
[3] https://issues.apache.org/jira/browse/COCOON-1424
Here is the diff between Vadim code and the
Joerg Heinicke escribió:
When looking at the history of ResourceReader I also noticed something
else:
In rev 155087 [1] Vadim did the following change:
Move response header initialization into the setup phase.
Remove Last-Modified header initialization: this is done in
the environment by request from AbstractProcessingPipeline.
In the next revision 410112 [2] (but 15 months later) Antonio readded
setting the Last-Modified header:
Fix COCOON-1840 XMLFileModule: file-specific configuration ignored.
The commit message seems to be wrong and the change was not included
in the user-provided patch [3]. It rather seems to be COCOON-1266 [4]
but the provided and applied patch is older than Vadim's change. I
guess we should review this code.
The revisions for trunk are 155099 [5] and 410113 [6].
Joerg
[1] http://svn.apache.org/viewvc?view=revrevision=155087
[2] http://svn.apache.org/viewvc?view=revrevision=410112
[3] https://issues.apache.org/jira/browse/COCOON-1840
[4] https://issues.apache.org/jira/browse/COCOON-1266
[5] http://svn.apache.org/viewvc?view=revrevision=155099
[6] http://svn.apache.org/viewvc?view=revrevision=410113