Moin

Bugfixes:
* Fixed MoinMoinBugs/XmlRpcBrokenForFastCgi - thanks to Johannes Berg.
* Fixed gui editor converter confusing of `` and {{{}}} markup.
* Fixed emission of HTTP headers (esp. Vary: Cache-Control:).
* Fixed a bad crash that happens (on ANY page!) when you put a *Dict
page's name as a member into a *Group page.
* Fix MyPages action title to use an unformatted string.
* Fix double quoted-printable encoding in generated emails (note: this is
not a moin bug - this just works around bugs in python stdlib).
* Fix mode of drawing files (use config.umask).
* Fix trouble with /?action= urls by dropping getPageNameFromQueryString
code.
* Fixed sre unpickle issues seen on some machines by avoiding to pickle
the regex.
* Fix Clock code, add more timers.
* Worked around FastCGI problem on Lighttpd: empty lines in the error log.
* Fix (add) locking for caching's .remove() call, small fixes to lock code.
* Print error message when --target-dir=path is missing from moin export
dump command.

New features:
* Added support for "304 not modified" response header for AttachFile get
and rss_rc actions - faster, less traffic, less load.
* Limit rss with full diffs to 5 items.
* Allow surge_action_limits = None to disable surge protection.
* moin.fcg improved - if you use FastCGI, you must use the new file:
* can self-terminate after some number of requests (default: -1, this
means "unlimited lifetime")
* the count of created threads is limited now (default: 5), you can use
1 to use non-threaded operation.
* configurable socket.listen() backlog (default: 5)
* Added indonesian i18n (id).
* Some measures against spammers trying to brute force their spam into moin.
* EmbedObject macro: added mov, mpg and midi support
* moin ... export dump --target-dir=PATH --page=PAGENAME_REGEX
You must specify --target-dir (where dump will write the html files to)
and you may specify --page and either a page name or a regex used to
match all pages you want to export. Default is to export all pages.

Other changes:
* Tuning:
* more efficient locking code on POSIX platforms, we do much less I/O
for locking now
* removed most chmod calls in favour of a single os.umask call
* Improved Python 2.5 compatibility. Note: if you think that MoinMoin
is using too much RAM, you might want to look at Python 2.5 because
of its improved memory management.
* Throw away SAVE comments longer than 200 chars (you can't enter those by
the user interface, so only spammer using automatic POSTs do them).
* Updated spider user agent list.
* BadContent and LocalBadContent now get noindex,nofollow robots header,
same as POSTs.
* Standalone opens it logfile unbuffered from now on, thanks to
Carsten Grohmann.
* Avoid trouble when request.write() data contains None, just skip it -
thanks to Juergen Hermann.
* Instead of showing a ConfigurationError, moin now emits "404 Not found"
http headers and a hint about what could be wrong. This won't fill up
your logs just because of typos and spiders still trying old URLs.

Bugfixes:
* fixed ticket creation to only use constant values

HINT: read docs/README.migration.
Version 1.5.4-current:
* increased maxlength of some input fields from 80 to 200

Version 1.5.current:
Developer notes:
* We switched to Mercurial SCM, see here for more infos:
http://moinmoin.wikiwikiweb.de/MoinDev/MercurialGuide

Bugfixes:
* fix MonthCalendar macro for non-ASCII pagenames (thanks to Jonas Smedegaard)
* remove 'search' and 'google' from bot user agent list and add some more
specific terms
* Fix the forgotten password email login URL and also properly encode the
email body. Thanks to Robin Dunn for the patch.
* Applied a patch by Matthew Gilbert for increased compatiblity with
latest docutils.

New Features:
* Added CSS classes for TableOfContents macro.

Bug Fixes:
* GUI editor / GUI editor converter:
* Fixed missing GUI editor link in edit bar.
* Fixed table/row/cell style spaces.
* Changed <p> generation for macros.
* Fixed handling of subpages.
* Fixed processing of complex list elements (thanks to Craig Markwardt).
* Fixed processing of html \n (thanks to Craig Markwardt).
* Fixed joining of comment lines with text below them.
* Fixed table height attribute crash.
* Fixed sslclientcert auth.
* Added some missing files to i18n processing, translatable strings more
complete now.
* Change <p> generation from self-closing (<p />) to auto-closing (only
emit <p>, let browser find place for </p>).
* Fix eating of newline before tables.
* Fixed incorrect login hint for not logged-in users trying a disallowed
action.
* Fixed nasty missing or double </a> formatter bug (mostly happening when
user has enabled questionmarks for nonexisting pages).

Other Changes:
* We catch and ignore html to wiki conversion errors in case of hitting
the cancel button, so you can get out of the error screen.

New Features:
* Modified SystemInfo macro to give human readable units and disk usage
* cfg.editor_quickhelp makes the quick help below the editor configurable
(at least as far as the default_markup is concerned). If set to None,
it doesn't display any quickhelp. Thanks to Seth Falcon for the patch.

Bugfixes:
* Fixed double class attribute for nonexistent links
* Fixed double </a> with qm_noexist option
* Fixed table xxx="yyy" style attribute parsing
* If not (editor_force and editor_default == 'text') then display GUI mode
switch button (this is only a partial fix, but enough to deny the GUI
mode to your users completely)
* Fixed XSS issue which could lead to cookie theft etc.
* Fixed definition list "same level" indenting.
* Fixed pagename in log for PackagePages action.
* Made <p> self-closing and line-anchors more well-behaved, thanks to
Martin Wilck for the patch. I didn't apply the <ol> patch, because this
is no real problem, just a problem of a bad standard.
* Fixed gui editor *bgcolor crash.
* Support/Fix tableclass attr with gui editor.

Other changes:
* Moved back UserPreferences action link from menu to top of page (renaming
it to "Preferences"), added "Cancel" button to make it possible to return
to the previous page without saving preferences.
* Removed formatter.url "attrs" keyword arg that took premade html, we use
separate kw args per attribute now.
* Moved broken tests to MoinMoin/_tests/broken (= disabling them).
All "active" tests run ok now.
If you want to compare, I included the output of this test run:
$ make test >tests/make_test.out

New features:
* HINT: please read README.migration
* Login and Logout are actions now, therefore you stay on the page where
you were before.
* UserPreferences is also an action now and moved from top border (in
modern theme) to the "more actions" dropdown menu. You also stay on the
same page.
* There is also a [[Login]] macro now. You can put it on any page you want
and if the user is not logged in, it renders the login form. If the user
is logged in, it doesn't render anything.
* We check cfg.superuser to be a list of user names (as documented) and
deny superuser access if it is not. This avoids security issues by
wrong configuration.
* auth methods changed:
HINT: if you wrote own auth methods, please change them as you see in
MoinMoin/auth.py and test them again before relying on them.
* now get a user_obj kw argument that is either a user object returned
from a previous auth method or None (if no user has been made up yet).
The auth method should either return a user object (if it has
determined one) or what it got as user_obj (being "passive") or None
(if it wants to "veto" some user even if a previous method already has
made up some user object).
* return value of continue_flag is now True in most cases (except if
it wants to "veto" and abort).
* moin_cookie auth method now logs out a user by deleting the cookie and
setting user_obj.valid = 0. This makes it possible to still get the
user's name in subsequent auth method calls within the same request.
* added ldap_login and smb_mount auth methods, see MoinMoin/auth.py and
wiki/config/more_samples/ldap_smb_farmconfig.py
* MonthCalendar now takes an additional argument for specifying a template
to use to directly invoke the page editor when clicking on non-existing
day pages.
* Added ImageLink macro. Thanks to Jeff Kunce, Marcin Zalewski, Reimar
Bauer and Chong-Dae Park for working on it.
* Lupy stuff (still experimental, partly broken and disabled by default):
* Attachment search using lupy (lupy_search = 1 in your config)
Title search will also search attachment filenames.
Full text search will also search attachment contents.
* Indexing filter plugins, see MoinMoin:FiltersForIndexing
Some filters depend on external converters that might not be available
for any platform (see Depends: line in filter plugin).
Feel free to contribute more filter plugins, especially if you wrote
some nice platform independant filter in Python for some popular file
format! Filters implemented so far (depending on):
MS Word, RTF, MS Excel (antiword, catdoc)
PDF (pdftotext)
OpenOffice.org XML based data formats (-)
XML, HTML (-)
text/* (-)
JPEG's EXIF data (-)
Binary generic (-)
* As you might know, Lupy is "retired" (that means it isn't developped
by its authors any more). We are currently evaluating Xapian as new
indexing search engine in moin.
If we succeed, we will replace Lupy stuff by some Xapian interface
code in moin.
But: the filters will likely stay, as we also need them with Xapian.

* A new MoinMoin script interface was introduced:

Syntax: moin [options] <cmdmodule> <cmdname> [options]

For example:
moin --config-dir=/my/cfgdir --wiki-url=wiki.example.org/ \
export dump --page=WikiSandBox

This will call the "moin" script, which will use the --config-dir and
--wiki-url options to initialize, then it will go to MoinMoin.script
module, import the export.dump module from there and run it, providing
the additional --page value to it.

The old scripts that have not been migrated to this new interface can
still be found in MoinMoin/script/old/ - including the old migration
scripts.
* moin ... account create --name=JoeDoe --email=joedoe.org
* moin ... account disable --name=JoeDoe
* moin ... acount check <-- this is what usercheck script was before
* moin ... maint cleancache <-- this is what cachecleaner script was
* moin ... maint cleanpage <-- this is what pagescleaner script was

Bugfixes:
* cookie_lifetime didn't work comfortable for low values. The cookie was
created once on login and never updated afterwards. So you got logged
out cookie_lifetime hours later, no matter whether you were active at
that time or not. This has been changed, we update the cookie expiry now
on every request, so it will expire cookie_lifetime after your last
request (not after login).
* Fixed logout problems when using cfg.cookie_path.
* Fixed cookie_path for root url wikis.
* Lupy search now behaves a bit less funky. Still no guarantees...
* We lowered the twisted server timeout to 2 minutes now (was: 10) because
it just used up too many files (open TCP connections until it timeouts)
on our farm.
* The algorithm used for processing the farmconfig.wikis list was changed
to work for big farms (>= 50 wikis), too. This works around the python
"re" module limitation of having a maximum of 100 named groups in a RE.
* Fixed a TypeError which occurred for formatters that dislike None values.
(cf. http://moinmoin.wikiwikiweb.de/MoinMoinBugs/PythonErrorEditingFrontPage)
* Fixed GUI editor converter error for https:... image URLs.
* ThemeBase (use by modern/rightsidebar): removed duplicate AttachFile from
actions menu (we already have it in editbar).
* Speedup group/dicts scanning for persistent servers.
* Implemented HEAD requests for standalone server, this should fix some of
the strange effects happening when using "Save as" on attachments.
* Not a moin bug, but rather a workaround for non-standard non-ASCII DNS
hostnames: we just use the IP instead of crashing now.
* Spurious cyclic usage error in i18n fixed.
* Fixed antispam for python 2.5a xmlrpclib.
* I18n for linenumber toggle in listings.
* All action menu entries are translatable now.

Other:
* Added css classes for the rst admonitions. Thanks to TiagoMacambira!