Release: Movable Type 3.2 Beta 4
This is an archived entry from the Movable Type 3.2 beta test.
Entries from the current beta release (v3.3) can be found here.
Beta-4 (tar/gzip, zip) is now ready and we're pretty sure that this is going to be the last one before the production release. It's just that good. We'll post the full release notes, changes and update the known issues list tomorrow, but I'll mention a few things now:
- Back up your databases!: See the new MySQL, SQLite, and BerkeleyDB instructions
- Make sure to read the installation/upgrade instructions if you haven't
- This release includes the new SpamLookup written for Movable Type 3.2. If you have altered your junk threshold in your feedback settings, please set it back to 0 initially. Using the defaults, it's working perfectly on all of our blogs.
- The help link in the top nav now points to the new help documentation
Bugs fixed in beta-4:
- No longer allowing ".." in Site URL because it's illegal in URLs (Case 12269)
- Fixed a problem with upgrade for a very limited set of users where it stops in the middle reporting "Error during upgrade: Error during upgrade: ." (Cases 12250, 12251, 12362, 12336, 12554)
- Fixed a commenter session timeout bug for authenticated commenters which casues either a "Registration is required" error or comments to be submitted as an unauthenticated commenter, depending on your feedback settings (Case 12615)
- Fixed an upgrade bug causes incomplete weekly archive list (Case 12246)
- Notifications are no longer sent for Junked comments (Case 12641)
- Fixed a bug with the "Entries to Display" setting, which if set to N entries, would display the first N entries posted to the weblog, rather than the last N entries. (Case 7224)
- Removed a non-existent tag (MTBlogIfDescription) from the Atom index (Case 12326)
- Fixed a bug where MT was not remembering login between page requests under mod_perl (Case 12355)
- Now shipping with a 3.2-compatible SpamLookup (Case 12109)
- Fixed a bug which caused MTCommentEntry to malfunction in dynamic category archives (Case 12356)
- Fixed a bug where <$MTEntryLink archive_type="Category"$> would throw error in individual archive template. (Case 12330)
- Fixed a bug where users lacking the Create Weblogs permission couldn't see a link to the system weblog list, even though the page itself would be safe to display (it is filtered to just accessible weblogs). (Case 12301)
- UpgradeScript config directive implemented now used by MT (Case 10110)
- Changing Author Username no longer results in forced logout (Case 7290)
- Fixed a bug where an entry title of '0' would cause basename to be set to 'post'. (Case 7013)
- Weblog activity log search no longer requires system level activity log permission (Case 7621)
- Fixed a bug where blog permissions of blog admin would become unchecked/enabled in IE after save (Case 12275)
- Fixed a bug which caused non-UTF8 encoded L10N lexcions to be broken in some random case (Case 12264)
- Fixed a bug where extra linefeeds in comment text weren't being stripped (Case 12249)
- Fixed some localization problems for certain scripts other than the admin scripts (Case 12248)
- mt-check.cgi is now localized (Case 12242)
- Fixed a bug suppressing the error message in the activity log when TypeKey auth signature validation failed (Case 7456)
- Templates with no name were uneditable because of lack of link text. Now prints "(no name)" (Case 12282)
- Default templates were not localized and some link title attributes were not being wrapped in MT_TRANS tags for translation (Case 12270, 11475)
- MTTopLevelParent no longer returns an error if NULL (Case 12181)
- Fixed the link to activity log for ping error on the edit entry page which caused it to open in a new window (Case 5563)
- Dropped RSS 1.0 template from the default set and added Atom 1.0 (Case 7379)
- Fixed a bug where password hint was not a required field on the profile editing screen (Case 8767)
- Fixed a number of display and information layout issues on the default templates (Case 12196)
- Switched to Windows linebreaks in the
mt-config.cgibecause they are universally compatible (Case 12307) - Fixed the misleading page title on Quickpost (Case 12337)
- Fixed an problem which caused an error to be thrown: "Can't call method 'clear_cache' on an undefined value at lib/MT/App.pm line 606." (Case 12525)
- Fixed a bug which caused category itemset actions to remain unpopulated. (Case 12648)
- Fixed a QuickPost bug in IE that would prevent blog selection (Case 12649)
- Fixed the out-of-date copy on the notification emailsdate (Case 12651)
- Fixed a bug in SpamLookup which would credit items with no links in the link memory test (Case 12652)
- Now giving a useful error when the SQLite DB or the directory which contains it is non-writeable (Case 12716)
- Fixed a bug in SpamLookup which would adversely affect pings sent from blogs with numeric IPs (Case 12823)
- Fixed a bug where untrusted commenters are able to publish comments when publish immediately from trusted commenters was selected (Case 12825)
- Fixed a bug where an ampersand in the category label would cause a double-underscore in dirification (Case 12826)
- RefreshTemplates is now preserving linked files with the old template (Case 12867)
- Removed the remember me from the preview page because it can be misleading in cases where MT domain and weblog domains differ (Case 12887)
- Now suppressing Save and Rebuild button in case of authors without rebuild permissions. (Case 12891)
- Fixed out-of-date breadcrumb in weblog settings page (Case 12942)
- Fixed the error thrown when a IP is submitted on the IP banning page (Case 12944)
- Fixed a bug which caused the nofollow plugin to create separate rel attributes in PHP versions before v4.2 (Case 13124)
- Fixed a problem on BerkeleyDB which would seem to disassociate TrackBacks with their entries on import (Case 13427)
- Default language is now localized in the config file (Case 13495)
- Fixed a problem where MT->version_number would fail if the version was not numeric (as is the case with any beta). Introduced VERSION_ID which holds the full alphanumeric version in addition to VERSION which holds the base numeric version. (Case 13505)
- Fixed a problem for those using advanced archive paths when creating uploaded image thumbnails (Case 13358)
Posted on August 12, 2005 8:03 PM in Announcements


Comments
Back up your databases!: See the new MySQL, SQLite, and BerkeleyDB instructions...
A curious omission there makes me wonder if the reason why I never got any replies to my semi-annual question about recentlycommentedon and postgres is because you're dropping support for postgres.
Are you?
Posted by: The One True b!X
|
August 12, 2005 9:36 PM
Not in the least! The honest truth is that I was the one who wrote up all of the backup instructions while everyone else was going a million miles an hour doing their jobs, and I don't know the first thing about Postgres. There's an entry in draft that says nothing more than:
???????????
I am going to have one of the engineers write it up on Monday. No conspiracy theories there. That's just how it goes with Beta docs. :-)
Posted by: Jay Allen
|
August 12, 2005 10:47 PM
Installed fresh and running sweet. I can't seem to break it. :)
Posted by: demonsurfer
|
August 13, 2005 12:05 AM
Running smoot. You guys must have done a lot of optimisation on the code, cos it's smokin :)
Posted by: Sara
|
August 13, 2005 12:24 AM
Is backing up using the export feature not enough anymore?
Posted by: Kai
|
August 13, 2005 6:57 AM
is the "offset" feature for MTEntries working for anyone else?
Posted by: Kevin Story
|
August 13, 2005 7:13 AM
The export feature is as it has always been. It backs up entries, comments and TrackBacks from a single blog. It doesn't preserve your configuration or templates.
Backing up your database preserves the whole installation so that you can roll back to exactly what you had when you backed up.
Posted by: Jay Allen
|
August 13, 2005 7:53 AM
I'm surprised it was considered a bug that notification was sent for junk email. I guess I don't get a lot of spam, but I liked knowing that I'd received junk so that I could check and make sure it really was junk. Now I'm afraid real comments will languish in my junk bin for days.
Great work on all this. I've been really pleasantly surprised at how few bugs even the nightly builds had.
Posted by: Matt Moore
|
August 13, 2005 8:45 AM
Thanks Matt. You wouldn't be so happy to receive notifications of junk if you got a lot. I get over 500 per day and was absolutely floored to log in and get all those emails. It reminded me of the early days before MT-Blacklist.
Posted by: Jay Allen
|
August 13, 2005 8:53 AM
That's what I figured. Maybe 3.3 can have a toggle button for junk notifications?
Posted by: Matt Moore
|
August 13, 2005 9:06 AM
I make no guarantee, but for Postgres:
pg_dump -U USERNAME DBNAME > FILENAME
That's the command I use to backup my database. I haven't had to restore from it yet, so I'm not positive it's correct, but a quick scan of the file looks fine to me.
Posted by: g026r
|
August 13, 2005 10:55 AM
Anyway I was just being cranky. It's one of those weekends.
FYI, if you run your postgres backups directly from the postgresql user on the system, I'm fairly certain you can use something like this:
/usr/local/pgsql/bin/pg_dump communique > /usr/local/pgsql/backups/communique.bak
(Same as above, but without the username argument. That's how I run my twice-daly DB backups anyway. And I did have to make use of them once when my Mom hosed her entire blog via the "search and replace" within the CMS.)
Posted by: The One True b!X
|
August 13, 2005 12:30 PM
Sigh. One of those weekends. That should have used placeholders for databsename and filename, but of course it didn't.
Posted by: The One True b!X
|
August 13, 2005 12:31 PM
Not that it's too important, but the default stylesheet comments say HOSE where I believe they out to say HOUSE.
This release fixed the problem I was having with not being able to manage comments in beta 2, so hoorah!
Posted by: Lauren
|
August 14, 2005 2:23 AM
Ack. You guys gotta check how the stylesheet and other templates are linked in the header and sidebar. Right now they're linked by filename instead of template name. Got a few extra grey hairs on that account ;P
Posted by: Sara
|
August 14, 2005 4:04 AM
I get follow message when 1st entry
Building entry 'blog4�エントリ' failed: Build error in template 'Individual Entry Archive': Error in tag: Can't find template 'styles-site.css'
Posted by: AS
|
August 14, 2005 7:35 AM
AS Is this a new blog from beta 4 with the dafault templates?
If the case, look through the templates and see if the stylesheet is linked correctly.
Posted by: Sara
|
August 14, 2005 9:00 AM
Hey Jay,
This beta fixed that thumbnail creation I was having with the archives folder! Weeeew!
Posted by: Kai
|
August 14, 2005 12:12 PM
I love the improvements you've made to Spam Lookup, especially the addition of credits for URLs or email addresses that have been published in previous comments, and giving specific demerits to specific keywords. You have enough user configurability here that the system might moderate most spam without any significant help from me. I appreciate that!
Posted by: Charlie
|
August 15, 2005 4:19 PM
I'm still using Beta 3, and have a quick question regarding dynamic templates. I never used this feature prior to installing Beta 3, and noticed that this feature seems to only work when updating the template within MT, and not when updating a linked template.
Is this a bug, or just a current limitation of the feature?
Thanks.
Posted by: phile
|
August 15, 2005 7:22 PM
This may be more of a support issue... I can "trust" certain commenters by selecting their comment and doing "More actions, trust commenter". Commenters can also be "authenticated" through Typekey. I can view a list of "authenticated" commenters, but I don't see a way to view a list of "trusted" commenters, and marking them as trusted does not make them show up on the authenticated list (MT seems to maintain a distinction between the two classes). How do I see who I've marked as trusted?
Posted by: Charlie
|
August 16, 2005 8:45 AM
Charlie: on the commenters listing, there is a "Filter:" line at the top. Click the hyperlink on the right of that to display the filter controls. You can change that to show only trusted commenters. You can also click on the "trusted" status icon of an existing trusted commenter to do the same filter.
Posted by: Brad Choate
|
August 16, 2005 9:29 AM
Ok, I see that. The trouble is that no matter how I filter, no commenters are shown on the commenter list. It's empty whether showing all or some filtered set. I assume the commenters list pulls its data from the list of comments... I can go to commenter A, select, choose to "trust" him, but his name doesn't appear on the list. I doubt this is a bug. Is there a setting that can prevent commenters from showing up?
Posted by: Charlie
|
August 16, 2005 11:39 AM
Hey Charlie, this sounds like a user support problem. You may want to post over here with links to screenshots of what you're talking about. I have a feeling that you're trying to trust unauthenticated commenters which, of course, doesn't work.
Posted by: Jay Allen
|
August 16, 2005 11:44 AM
Thanks. I'll do it. Why is trusting unauthenticated commenters an "of course it doesn't work" issue? Does that mean I can't choose to trust someone if they haven't been authenticated by Typekey?
Posted by: Charlie
|
August 16, 2005 11:48 AM
Because it's based on authenticated identity. Authenticated ID is mapped to a person. A name and email address is mapped to letters on a screen that anyone can type in. There are (and will be) plugins available for whitelisting and blacklisting commenters by such random and mutable properties like their Name or Email Address, but that's not a solid foundation for "Trust".
Posted by: Jay Allen
|
August 16, 2005 11:53 AM
That helps. As Reagan famously said, "Trust, but verify."
Posted by: Charlie
|
August 16, 2005 12:05 PM
Hi Jay, I'm encountering the same issue as Charlie. I think this is a UI problem. It isn't clear that "Trusted" is a subset of "Authenticated. In fact, on my MT 3.2 blog I have "Immediately publish comments from" set up for "trusted commenters only". I have not set up any authentication, because I didn't think I needed any.
The result is that one someone goes to comment, they get the following message: "I have enabled a feature that allows your comments to be held for approval the first time you post a comment." The problem is that their comments always have to be approved, because I haven't enabled authentication.
I suggest graying out the "Trusted" box on the feedback settings until the authenticaiton is selected and token placed.
Posted by: elise
|
August 16, 2005 2:03 PM