| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
| |
No more %-strings. Kill off all __i18n_templates__ hacks.
|
| | |
|
| |
|
|
| |
password for the user, or if it's "{NONE}".
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
column in the database for this list of strings. We use SQLAlchemy's
many-to-many relationship, however because of this, you cannot simply append
new unicodes to .available_languages. You need to wrap the language code in a
Language instance and append that instance to the list.
In order to handle this, I added a property MailList.language_codes which
returns a list of the code strings (not Language instances). Also new are
MailList.set_languages() for setting (i.e. overriding) the set of available
languages for the list; and add_language() which takes a single language code,
wraps it, and appends it. The code does not and should not use
.available_languages directory any more.
MailList.GetAvailableLanguages() is removed. The 'available_languages' column
is removed from the Listdata table.
Add a getValue() to Mailman.Gui.Language in order to unwrap the language codes
stored in the database's association table. Modify _setValue() to do the
wrapping.
In dbcontext.py, don't import * from the sqlalchemy package. It contains a
'logging' name which is not the standard Python logging package. I also added
essentially a bag of attributes class called Tables which will hold references
to all the SA tables that are created. Update the make_table() API to take an
instance of Tables.
Added a close() method to DBContext. This is needed for the updated unit test
suite.
Changed bin/import.py so that when available_languages is being set, it calls
MailList.set_languages() instead of trying to set that attribute directly.
Updated some language idioms while I was at it.
More eradication of mm_cfg in favor of the config object and the Defaults
module.
In testall.py, call initialize() instead of loginit.initialize().
Promote MAX_RESTARTS into a Defaults.py.in variable. This is because the unit
tests will knock that value down to something not so annoying should one of
the qrunner-required tests traceback.
Several other important changes to the unit test suite (which now completely
succeeds again!):
- Set the uid and gid of the temporary mailman.cfg and tmp*.db files to the
Mailman user and group as specified in the config object.
- Make sure that all of the tests point to a SQLite database file that was
created with the tempfile module. This way we don't pollute our main
database with data that is getting created during the unit tests.
- In the TestBase.setUp() method, be sure to close the existing dbcontext,
clear out the mappers, and then reconnect the dbcontext with the new
SQLALCHEMY_ENGINE_URL pointing to the tempfile. However, we don't need to
reload the MailList instance any more.
- Make all tests work, except for the tests that require crypt. That upgrade
path will not be available in this version of Mailman.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
First off, there are several password hashing schemes added including SHA,
salted-SHA, and RFC 2989 PBKDF2 (contributed by Bob Fleck). Then we encode
the password using RFC 2307 style syntax. At least I think: specifically
things like the PRF and iteration count for PBKDF2 are encoded the way I
/think/ is intended for RFC 2307 but I could be wrong. Seems darn hard to
find definitive information about that.
In any event, even though CLEARTEXT passwords are supported, they are mostly
deprecated, even for user passwords. It also allows us to easily update all
passwords to a new hashing scheme when the existing schemes get cracked. The
default scheme (specified in Defaults.py.in) is salted-SHA with a 20 byte salt
(the salt length and PBKDF2 iteration counts can only be specified in the
passwords.py file).
These hashed passwords are used for user passwords, list owner and moderator
passwords, and site and list creator passwords.
Of course this means that user password reminders are impossible now. They've
been ripped out of the code for a while, but now we'll need to implement
password resets since user passwords cannot be recovered.
bin/export has had several changes:
- export no longer converts to dollar strings. Were assuming dollar strings
are used by default for all new lists and any imported lists will already be
converted to dollar strings.
- Likewise, rip out the password scheme stuff, since cleartext passwords can
never be exported, so we might as well always include the member's hashed
password.
- Fix exporting to stdout when that stream can only handle ascii by wrapping
stdout in a utf-8 codec writer.
Other changes:
- add a missing import to HTTPRunner.py
- Convert GUIBase.py to use Defaults.* for constants instead of mm_cfg.*
- Remove pre-Python 2.4 compatibility from Utils.py. We've already said
Python 2.4 will be a minimum requirement.
- Change the permissions on the global password file. The default 007 umask
is used and should be good enough.
- bin/newlist adds the ability to specify the password scheme (or list the
available schemes) for the list owner password. It is not possible to set
the scheme on a per-list basis. bin/mmsitepass does the same, but for the
site and list creator passwords.
- Fix a nasty problem with bin/import. The comment in the code says it best:
# XXX Here's what sucks. Some properties need to have
# _setValue() called on the gui component, because those
# methods do some pre-processing on the values before they're
# applied to the MailList instance. But we don't have a good
# way to find a category and sub-category that a particular
# property belongs to. Plus this will probably change. So
# for now, we'll just hard code the extra post-processing
# here. The good news is that not all _setValue() munging
# needs to be done -- for example, we've already converted
# everything to dollar strings.
- Set the 'debug' logger to logging.DEBUG level. It doesn't seem to make much
sense for the debugging log to ignore debug messages.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
during early initialization so that we're guaranteed to get the right value
regardless of the shell umask used to invoke the command line script. While
we're at it, we can remove almost all individual umask settings previously in
the code, and make file permissions consistently -rw-rw---- (IOW, files are no
longer other readable).
The only subsystem that wasn't changed was the archiver, because it uses its
own umask settings to ensure that private archives have the proper
permissions. Eventually we'll mess with this, but if it ain't broken...
Note that check_perms complains about directory permissions, but I think
check_perms can be fixed (or perhaps, even removed?!). If we decide to use
LMTPRunner and HTTPRunner exclusively then no outside process will be touching
our files potentially with the incorrect permissions, umask, owner, or group.
If we control all of our own touch points then I think we can lock out
'other'.
Another open question is whether Utils.set_global_password() can have its
umask setting removed. It locks permissions down so even the group can't
write to the site password file, but the default umask of 007 might be good
enough even for this file.
Utils.makedirs() now takes an optional mode argument, which defaults to 02775
for backward compatibility. First, the default mode can probably be changed
to 02770 (see above). Second, all code that was tweaking the umask in order
to do a platform compatible os.mkdir() has now been refactored to use
Utils.makedirs().
Another tricky thing was getting SQLite via SQLAlchemy to create its
data/mailman.db file with the proper permissions. From the comment in
dbcontext.py:
# XXX By design of SQLite, database file creation does not honor
# umask. See their ticket #1193:
# http://www.sqlite.org/cvstrac/tktview?tn=1193,31
More details in that file, but the work around is to essentially 'touch' the
database file if 'sqlite' is the scheme of the SQLAlchemy URL. This little
pre-touch sets the right umask honoring permission and won't hurt if the file
already exists. SQLite will happily keep the existing permissions, and in
fact that ticket referenced above recommends doing things this way.
In the Mailman.database.initialize(), create a global lock that prevents more
than one process from entering this init function at the same time. It's
probably not strictly necessary given that I believe all the operations in
dbcontext.connect() are multi-processing safe, but it also doesn't seem to
hurt and prevents race conditions regardless of the database's own
safeguards (or lack thereof).
Make sure nightly_gzip.py calls initialize().
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Ensure that exported XML is written in utf-8, at least if we're writing to
a file other than stdout. Fix a typo in getting the digest style. Update
copyright years.
In HTTPRunner.py, catch KeyboardInterrupt. In Python 2.5 this has been moved
in the exception hierarchy so that it's no longer caught by "except
Exception".
import.py: Import user topic selections. Fix a typo in an error message.
Catch BadDomainSpecificationErrors that can be raised in MailList.Create().
|
| | |
|
|
|
header_filters and topics (both list topics and user topic selections).
Everything else seems to work pretty well.
dbcontext.py: Don't key the mlist transactions off of mlist.fqdn_listname
because this can change. For example, if you "bin/withlist -l mylist" and
then "m.host_name = 'new.example.com'" the fqdn_listname property will change
and the commit machinery won't be able to find the correct transaction.
Instead, store the fqdn_listname as it's seen during the api_lock() call back
on the mailing list under the _txnkey attribute. Use that attribute in
api_save() and api_unlock().
Upgrade to SQLAlchemy 0.3.3
Port from MM2.1 the support for multiple password schemes.
Change the MailList's repr to use the fqdn_listname.
|
