Revision history for Apache2-Controller
1.001.001 2014-06-11
Was the UNIVERSAL bug fixed in the last version? It looked
that way.
Corrected some POD errors.
t/render.t was failing because I had set the relative_uri
with the uri if there was no relative_uri. That didn't
make sense, it resulted in a url /foo/default/default.html.
All the other tests seem to pass still. I don't know why
I introduced that bug or why it appeared at the time
that the test worked with that regression.
Look for A2C in Debian!
1.001.000 2012-07-??
Net::OpenID::Consumer & ::Server required to build in 5.12,
even though the OpenID module doesn't work yet. I really
need to get to figuring that out, but I don't have a use case.
A2C_Render_Template_Path directive needed assignment of
empty arrayref if not specified but used somewhere.
Dispatch::Simple assigns the requested uri if no matching
uri found, to kick back up to "not found" Apache errors.
Logging a bad request shouldn't limit reason length (as much).
It will get cut off anyway by whatever it sends it to,
however Apache logging is configured.
Render::Template now handles error status codes correctly,
returning DONE in the case an error occured and an error
template was used, so Apache doesn't try to append its own
html error messages to the finished html output.
Session factored tying and will now generate a new cookie
and session if the session store object referenced by the old
cookie is no longer present.
1.000.111 2010-12-21
bad cookie eval only if directive A2C_Skip_Bogus_Cookies.
1.000.101 2010-12-21
mkdir explicitly uses $_ for perl < 5.10. Fixes RT# 60240.
Errors from Apache2::Cookie::Jar->new() skipped if the
APR::Request::Error code is NOTOKEN. Bad cookie, buggy
client. Fixes RT# 61744.
1.000.100 2009-02-02
Fixed MANIFEST. RT# 42979.
https://rt.cpan.org/Ticket/Display.html?id=42979
Switched to Module::Build to better control prereqs,
build prereqs etc.
Render::Template doesn't assign stash functions anymore,
in order to reduce number of dependencies. Do this yourself
by overloading render().
Test suite uses File::Temp tempdir now so that temp
directories get cleaned up properly.
No longer tries `od` with open3(), so MS Win might work?
A2C stash data is put in sub-hash 'a2c' of stash,
so it doesn't clutter the namespace and is consistent
with the same change that was made to consolidate pnotes.
1.000.011 2009-01-10
No more setting status_line() with HTTP::Status messages
for error conditions - just let Apache set its message.
Re-arranged a little bit of error handling so you can
set or return a status from error().
Bugfix for Methods get_cookie_jar - it should not store
a reference in pnotes because it contains a ref to the
Request.
Pushed Session handler now saves a top-level timestamp
to force update of Apache::Session, if you set the
directive flag.
Session save handler is now its own package,
concerned about closures keeping the request around.
Is this a problem with the DBI rollback handler,
which closes on a string var in the setup phase?
Moved all of the internal notes/pnotes garbage
under one structure $r->pnotes->{a2c} because
it makes more sense to not clutter the limited
notes/pnotes key namespace.
1.000.010 2008-12-2
Detaints directive values.
Controller doesn't double-check allowed_methods
in case your dispatch wants freedom from that.
Improvements/correction to documentation.
Session now uses a checkum. Thanks David Ihern.
OpenID auth module put on backburner until
I rewrite it again, I wanted to get this out
the door with functional session checksum and
correct documentation.
1.000.001 2008-12-20
Put CPAN upload in the right directory. Duh.
1.000.000 2008-12-20
Session now saves only if status < 300, or if
notes->{a2c_session_force_save} is set.
OpenID! Rockin. It doesn't preserve post or get vars
yet from the prior request across the openid redirect cycle,
but it's good enough to release. Hurrah!
Instead of using PerlCleanupHandler to save session and
commit database, it uses PerlLogHandler while the
connection is still open. Otherwise, session untying
and database commits may not be processed synchronously
before the next request that depends on those changes.
0.110.000 2008-08-06
Removing the "RUN_ALL" simulated mode from response handler.
It's an over-optimization, and A2C:DBI::Connector supports
a PerlCleanupHandler to rollback uncommitted transactions.
Now Controller.pm does not return DONE, but sets status
and returns OK.
Apache2::Controller::DBI::Connector works. Cool!
0.101.111 2008-07-22T18:01:00
LAME LAME LAME I hate versions.
0.101.110 2008-07-22T18:00:00
Oops I see, Version.pm was not in the MANIFEST, no wonder.
0.101.101 2008-07-22T02:30:00
Version numbers are so strange.
0.4.3 2008-07-21T00:22:00
Removed ability to limit methods. Use Apache config for that.
A2C:SQL::Controller. untested. Controller now passes @path_args
as arguments to method. Version is now in its own package.
Some new SQL-related packages that may or may not work.
Universal use of Apache2::Controller::Directives - no more
PerlSetVar. Documentation edits.
0.4.1 2008-07-16T00:16:00
Changing version number to use 'version' package.
I had the version number screwed up before.
The numbers below are actually 0.4.0, 0.3.0 etc.
0.0.4 2008-07-16T00:08:00
New dispatch method HashTree implements alternative dispatching.
Various improvements to documentation. More tests.
Developed unit test infrastructure.
0.0.3 2008-07-06T00:08:00
Changed render to always use A2CRenderTemplateDir as INCLUDE_PATH.
Various and sundry updates. Trying to get Directives to work.
Bleah.
0.0.2 2008-07-02T00:08:00
Woo-hoo! Rendering and sessions work. More tests.
0.0.1 2008-06-01T00:00:00
First version, released on an unsuspecting world.