Mcavity

Variable Interpolation

2010-07-11T14:35:38Z

MovableType has this incredibly beautiful thing called variable interpolation. As I understand it, what this means is that you can define a variable somewhere and then reference it at another place altogether. You can even use it to supply a value for a template tag modifier, another variable, or a value for a template tag if the tag takes one.

]]> Let’s look at an example, used on this very site, its siblings and parent. Here’s the set up, all on a single box at the same webhost. A single MT installation. There’s the parent site, gautampatel.com. Beneath that there are four existing blogs, including this one, and more are planned. The parent site has no entries of its own; it only pulls content from the subsidiary blogs. Each sub-blog has its own distinctive URL, too, a full-fledged domain name correctly mapped.

Thus:

Each blog, including the parent is individually styled. The styles of the sub-blogs do not reflect on the parent blog.

Evidently, each blog has its own URL, path and blogid. I need to identify all of these separately and then cobble all the ids together, separated by commas, so I can use Multiblog and Search effectively.

Up in the parent blog, I create a template module called System Head. It looks like this.

Here’s a line-by-line explanation:

This sets the Multiblog block to look at all the blogs in the system.

This uses a logic tag, if, and, as Multiblog cycles through the blogs, the logic tag tests whether a certain condition is true using a simple regular expression.

Next, if the condition is true, use the setvar tag to set a bunch of variables: the blogid, the blogname, the blogurl. There’s an easier way of doing this in one go, using ; I’ve just stuck with this.

Then you repeat the condition test and setting of variables for all blogs in the system.

Now, for the search and other things, I want a listing of all blogs except the parent blog. So I do this.

Note the variable interpolation. I used $GPID to pull the blogid of the parent blog from the variable I set earlier, and I excluded that. By the way, as far as I know, is the same thing as .

I also used here because I was doing this that the tag cannot do; I was pulling a value from another tag.

Note: Be careful using . You should have everything between the opening and closing tags in one line, because otherwise it inserts linefeeds (naturally; it’s doing a block). Alternatively, you can use the strip_linefeeds="1" modifier, so:

What this block does is that it loops through all the blogs in the system except the parent blog, pulls their blogids, and stashes them in an array (that’s the function="push" thing you see there. Again, note the variable interpolation.

Here’s the fun part. I now call that array, using the tag, get the values for each item in the array and stock them into another array.

The next lines are a redundancy. Just laziness.

Now I need to use all or any of these variable anywhere in the system. So what I do is that I link the System Head template module to a file on my system. Something like mod_syshead.mtml or whatever. Then, in the each sub-blog, I recreate a copy of the System Head module (pointing it to the file’s original location; I don’t make multiple copies of these files); and then add a line to the top of the HTML Head template module — it has to be right at the top, by the way — invoking the System Head template module.

So in the Mcavity sub-blog, the HTML Head template module has this line at the very top:

That way each sub-blog operates with the same set of variables. And I can use these variables to identify which blogs should be searched, or to use a variable like GPURL to reference a location under the parent website, like so:

    /images/mynicepic.jpg

Whew! Long explanation for something so seemingly trivial, but it’s not really. It’s logically true but also dense, and without MT’s marvellous variable interpolation, logic and array functions it would have been a nightmare.

Here’s the beauty and power of MT: It allows you to do so much, so elegantly and quickly with just a little patience and thought.

Tip of the hat to whoever thought of this.

Forum Non Convenience

2010-07-11T09:24:01Z

I’ve said this before, and I’m not the only one: the MT Forums is the clunkiest, most bizarre, user-unfriendly community service conceivable. To call it a ‘service’ is to give it credit it does little to earn.

]]> First of all, sometimes it uses Markdown. Sometimes it doesn’t. It’s not easy to figure out how to do links. It’s even tougher to figure out how to push out code (use Markdown; indent by eight spaces or more; but do you convert the < and > into <s and >s or not? Who knows?). There’s no preview for an originating post or new topic — take your chances, chum.

The search is ridiculous. Enter a term and you get a long, long list. At the bottom there’s an indication of more pages. There aren’t any. You can’t search by author. You can’t filter by date. There’s no concept of anyone moderating the forums or putting things where they should be so everything is everywhere.

It gets worse. Hit archives and you get a listing — by date. Duh? WTF? Who the heck is going to trawl through years of monthly archives to find something you need?

Oh, yes. There’s no spam control. None. Zip. Zilch. De nada. But there is an excuse, oops, explanation. Something about people using real accounts and therefore being trusted commenters by default. Here’s the thing about defaults: they can be changed. Anything else?

How do you edit a previous post, something that’s easily done on anything run on phpBB? You can’t. You have to log in. Then you have the pleasure of scrolling through and endless list (use the filters). You can also write a new entry here — so you don’t actually have to go the forums to post. What?

And it’s populated by some of the rudest people around, people whose approach is to say something like “Well buddy, nobody held a gun to your head, and besides we’re doing this free so show a little appreciation and bugger off” or things like that.

Hello? I came here looking for help. What do I find? A frustrating interface that, in and of itself, tells me that I’m insignificant and my worries, concerns and questions more so.

To call that a service or a forum is to do violence to the language.

Death by Calendar: The Irrelevance of Date-Based Blogs

2010-07-11T04:12:09Z

The historical fascination with “date-based” entries continues to invade all blog systems and software and many CMS platforms. Even today, every version of MT has provisioning for calendars and archives by date.

]]> This is just stupid. Content matters, not when it was written. The traditional view of blogs as personal online diaries is outdated. It is also fundamentally wrong.

If a blog is intended for a readership beyond the author himself, then it stands to reason that listing archives of blog entries by date is wholly illogical. A visitor to a blog does not know when the author wrote something. It makes no difference whether a particular subject was addressed in 2009 or 1995, or on what date of what month. Readers are interested in the subjects or topics being addressed. Category listings make sense; so do tag clouds, keyword listing and even alphabetical listings. And, of course, the search box is essential.

Many blogs continue to list date-listings in their archives. This is utterly useless. It is unthinking, uncaring and shows no respect for the reader or the visitor. Who cares if you wrote 200 blog posts in December of 2007? What did you write about? How do I find that?

The worthies at MT say that entries are date-based and pages are not. Entries have categories, pages have folders. This is a distinction without a difference. It is the kind of thing engineers push out because they’re not expected to think about design or usability or end-user convenience.

When you create a blog with entries and categories, the categories show up as — guess what — folders on your server drive. Page folders, ditto, naturally. So what’s the difference? Only this: entries default to that mindless date-based listing, while pages do not.

Then there’s this bit of nonsense: pages are static, stuff like an About or Contact pages, material that doesn’t change. Entries don’t much change either. You post one, then you move on. You might reference back to the previous one, but you certainly don’t keep changing an earlier post.

In MT there is no longer any justification for date-based archives. It’s one thing to have it there as an option; it’s quite an another to have it as a default archive mapping.

Wordpress.com, in its free version, is a great service, but for this: it will not permit you to change the default mapping. Entries should be listed by

    %category%/%postname%

or whatever syntax WP uses. Instead, WP forces entries to be listed by year/month/date. Get real, people: the date, month and year are irrelevant. Entirely.

If you’re at all serious about your blog, do this first: kill all date-based archives. Nobody’s interested.

So long, MT5. Sorry it didn't work out

2010-07-10T14:35:38Z

These things are never easy, but you can’t delay them indefinitely. For months I’ve struggled with MT5. It works just fine — up to a point. If it’s just the basics you want, and if you’re happy to live without almost every bell and whistle, and if you can wrap your head around the completely rejigged workflow and architecture, go ahead.

]]> But I love my bells and whistles, every one of them, some more than others. More: I need some of these plugins and hacks and I’m so dependent on them that they have now become the first point of reference when I check another CMS. Does it have anything like Dan Wolfgang’s Better File Uploader? The Asset Handler? Image Cropper? How does it deal with Markdown? Can I use PHP Markdown Extra? Collate entries, compare dates?

It’s the reason I’ve stayed away from shifting to WordPress or ExpressionEngine. I love the MT templating language; it’s intuitive, reasonably easy, and very flexible. Its great power comes from its native logic and variable template tags, most especially that near-magical thing called variable interpolation.

MT5 was a dream. I was entranced by its paradigm: simple, elegant, logical, centred around the MultiBlog principle of collecting stuff from different places and displaying them in one top-level overarching container unit. Therefore: the Website with its pages but no entries and, below that, Blogs with pages or entries; and the Website pulling content from blogs.

The dream shattered. It released very late, in January 2010 after months of waiting. The documentation was horrendous. Particularly acrimonious posts on the forums (arguably the single most ungainly, ill-designed and user-unfriendly such resources anywhere) resulted in the documentation being updated.

In all this time, SA USA seemed to have abandoned its constituency altogether: us. Only Japan seemed to flourish. I can’t read Japanese. I shouldn’t be expected to. And only Jun Kuneko from Japan came through again and again.

A word about Jun. The gentlest and kindest of men with the most exquisite good manners, he — and he alone — took all the harsh words (and there were many, on ProNet, on the forums) and responded. Not reacted. Responded. As in came through with documentation updates, patient explanations, detailed expositions, often in the face of a serious language barrier. SA needs to acknowledge the worth of this money and put him very high in the organisation. With him, MT is in safe hands.

And there’s the beleaguered Beau Smith, man with more interests, hobbies, occupations and passions than could possible fit into the largest NorthFace pack. In the face of the nearly incessant drubbing MT5 got, Beau kept his cool and stood by the product. Over on FogBugz, some of his questions in the beta-days of MT5 showed his profound concern for the end-user. Besides, by all accounts, he’s a master chef.

On his site, Beau describes SA as “awesome”. That’s now became an escape word, something that you use when you can’t, or don’t want to, more precisely say what it is you like or admire about the object of the term. SA isn’t awesome, not by a long shot. There was a time when SA seemed actually to care about its users. That’s simply not true of English-speaking users any more for any version of MT. SA has shifted its attention to TypePad and other products. MT fell by the wayside. I have no doubt that someone — and I have a fairly good idea who — is going to wade in with a tiresome tirade about how good and fine SA is. Stuff it. Look at the posts on ProNet over the past year. People have been saying that no one seems to be listening and at least one, possibly two, of the SA team, once agreed on Pronet that SA should listen and interact more with the community. That has not happened. SA’s own website is incredibly ugly. Blogs.com is even worse. Both seem tired, jaded, bored. The forums, as I said, are so clunky it’s unbelievable — a worse plug for the product is hard to imagine. And then there’s that continuing confusion between MT-org and MT-com and which has what and why, and the plugins are somewhere else. In the early days MT had the best documentation in the business, stuff you could download, beautifully indexed and cross-linked. Now it’s all over the place. You won’t find the documentation on variable interpolation in, or even linked in, the documentation on variable tags. Sometimes it’s in a comment. The real stuff is on another page altogether. Why? And there’s all kinds of hidden stuff, too, including modifiers for counts (which got documented very late in the day after Mihai Boscaru and I pointed to them) and more. What does this say about usability? Worse: what does it say about approach and attitude?

I was once a great apologist for MT5. I loved its logic. I admired what it was trying to do. I loved the elegance and simplicity with which content could be pulled from different sources and reshaped, repackaged and re-presented.

But intentions, however good and noble, don’t make a great product. MT5 sucks. Big time. Every single valuable MT4 plugin is borked on MT5 — at least all those that matter to me. Dan explained in an email and later on a forum post that plugin developers had been taken into confidence about the changes being made in the code during the shift to MT5. Often the plugins needs only small tweaks to get by: I managed to get Dan’s BFU running on MT5 with just a small change but it took hours of time; time I could have better spent elsewhere; time Dan wasn’t willing to invest (and rightly so); and it only ever ran in a crippled fashion.

Other plugins I had to toss altogether: notably the Asset Handler, which can be a life-saver if you’re running a parallel Xampp install, and Image Cropper. To be fair, lots of the legacy plugins did work, but the ones that didn’t made it a frustrating experience.

What finally did it was this: MT5 is totally rigid and inflexible in its URL structure between the so-called “website” and subsidiary blogs. The blogs cannot have their own independent URLs. You cannot change that setting. An exchange on FogBugz points to this, and says it should have been thought of, but it never was.

Here’s what I mean: the principal site here is gautampatel.com. Hierarchically below that, but on the same Bluehost box, are my blogs, each in its own directory/folder. This one, mcavity, is one such. Each of these blogs has its own domain name, correctly mapped to the Bluehost nameservers. Thus:

    www.mcavity.com

is the same as

    www.gautampatel.com/mcavity/.

On MT5, I cannot use the blog’s own URL. Everything has to go to through www.gautampatel.com, and every entry here has to be

    www.gautampatel.com/mcavity/category/entry.php

That’s just ridiculous. I want a system where I can post to any of my other blogs, have them appear under their own URLs, but it also reflects automatically on the main website, which correctly links back to the post on the sub-blog.

I moved back to MT4. It wasn’t easy. MT5’s multiblogging system and its new website-specific template tags make it a snap to identify sub-blogs and pull content. In MT4, everything is a blog, so you have to correctly identify each blog by its id and carefully map the URLs and blog names to those ids when you’re referencing them elsewhere.

Thank god for variable interpolation and the array functions. I managed to work out a system where from the main faux-website (faux because it’s actually an MT4 blog disguised as an MT5 ‘website’), I can do a search across all subsidiary blogs. I’ll push that bit of code out separately; I believe it should be of interest.

So sayonara MT5. It was good knowing you, but I guess this is the end of the road for us. I’m staying with my first love.

Sorting Categories in Movable Type: Three Solutions

2007-06-14T09:30:12Z

If you want to sort your categories in some customized order — i.e., not alphabetically, which is the MT default — there are some solutions. I’ve tracked down three different solutions.

(Note that I am now running MT 3.35, and I’m not sure Solution #3 below works in anything below version 3.3x.)

Use the RegEx Plugin;
Use the CustomFields Plugin;
Use category basenames.

]]> Option 1 : Use the RegEx Plugin

Here’s what you do:

Grab the plugin (here), and install it. If that link is broken, I’ve got the same thing locally, here.
Rename your categories 1FirstCat, 2SecondCat, 3ThirdCat. Note that you can also do 01FirstCat, 02SecondCat, etc. This may actually be preferrable, given the way these things are sorted if you have more than 10 categories. In that situation, the sequence will be 1, 10, 2, 3 …; while, with the double-digit numbering, it goes correctly 01, 02, 03.
Somewhere in your code, add this:
```
    s|^([\d\.]*\s*)||   
```
Now, when you use the tag, add the following argument:
```
<$MTCategoryLabel regex="RemoveSortNumber"$>
```
Basically, this is stripping digits from the start of the line.

Pros

Easy to implement
Requires minimal ‘treatment’ of categories

Cons

Does not work with dynamic publishing — returns a smarty error.
You have to know your way around regular expressions to modify the code.

Note that much the same thing can be done with the superb MT-Macros plugin. That’s really a three-and-halfth solution, because, as I see it, it’s only a refinement of the RegEx solution.

Option 2 : Use Arvind Satyanarayan’s CustomFields Plugin

Grab the plugin (here), and install it. Use the latest release, 2.0 RC1. This will not work with the earlier release.
Add a custom category field. Call it something like CategoryNum or some such. Check the required box — this prevents creation of a category without filling in a category number. The category type should be a single line text field so that you can add categories easily later.
I’m not going to get into how you get the custom fields to appear — Arvind’s documentation is exemplary, and bears close study.
Note this, though: the custom category fields do not change the entry editing screen, but the category creation screens.
Now when you create a category you should see a box at the bottom with your custom category. Give it a number (something like 01, 02, etc), in the sequence in which you want your categories to appear.
Now here’s the trick. Over at his own plugin forums at Movalog, there’s a thread where Arvind tells of an undocumented sort feature in Version 2.0RC1 of CustomFields. You can add an argument to the MTCategories tag and force a sort by a custom field. Here’s what Arvind says:

this is a new feature (that I haven’t publiciszed or documented at all for lack of time) introduced with v2.0. To resort by a custom field (called “Foo Bar”), the following syntax is used (on MTEntries, MTCategories or MTEntries):
```
       sort_by="CUSTOMFIELD:Foo Bar"
            sort_order="ascend|descend"
```
So supposing you wanted to resort categories by the Foo Bar field in descending order:
```
       
```

So in our case, with the CustomField CategoryNum, the code would be:

    
        
            
                <$MTCategoryLabel$>
                <$MTCategoryDescription$>

Essentially, all you are doing is sorting by the custom field, and then displaying as normal.

Pretty nifty.

Pros

Very easy, flexible, endless possibilities
Doesn’t require processing of a string.
Demands no special knowledge of reg-ex syntax.

Cons

None that I can tell, but it is a Release Candidate and — so the discussions on the plugin forums say — sometimes quirky. I’ve had no problems whatever.

Option 3 : Use the category to sort

This one’s sort of homegrown. Here’s what you do.

Name your categories as in Option 1 above, 01FirstCat, 02SecondCat, etc.
In the category editing screen, unlock the category basename field. It will default to the category label and has the numbers you’ve just given them (if you’re starting from scratch, that is; otherwise skip this step, the basename won’t change).
Strip the numbers from the basename and lock the field.
Remember to change your archive file path specifiers and mappings in the Settings - Publishing - Archive Maps so that your category archives are built with the basename rather than the label.

Now, instead of using the tag, you use the tag, passing it through any formatting filters (for upper/lower case, etc). For example:

    
        
            
                ">
                <$MTCategoryBasename$>

Basically, you’ve swapped the category label and the category basename. Now MT sorts categories by the label, in the numbered sequence you’ve set up, but uses the basename of each category for display.

Pros

Simple and basic. Requires no plugins, downloads, installations
Publishing-mode neutral; it makes no difference whether your pages are static or dynamic. MT is just being forced to read another field altogether.

Cons

Inherently limited and restrictive — it forces you to a given syntax in your basenames and labels. The basename can’t have spaces, just underscores and numbers. So if you have a multi-word category label ("01 My Favourite Things") the basename, after stripping the numbers, would be my_favourite_things. That’s really ugly. So you’d have to either limit yourself to one word categories, or run MT-Macros or the Regex plugin to get rid of the underscores.
Requires fiddling the archive mapping and file path specifiers.
Horrendously inelegant as a solution.

My choice? Option 2, without a doubt. Arvind rules!

Recipe for Upgrading MT

2007-06-10T16:22:53Z

After years of doing it donkey-style, I finally figured the easiest and fastest way to upgrade MovableType. I was moving from Version 3.33 to 3.35, so all references are to that. Here’s my very own Martha Stewart style upgrade recipe …

]]> Preparation Time: 5-10 minutes; Cooking time: 5 minutes, or depends on you. Serves as many as you like.

Ingredients

One working control panel to access your site (my host, Bluehost gives me the popular CPanel, so I’m presuming you have that).
A download of the latest MT version to which you want to upgrade.
A good zipping/compressing utility, one that creates tarballs (.tar) and gzip (.gz) files. You’re going to need a .tar.gz file. I’m sure this works just as well with a .zip file; I’ve just never tried it. I recommend 7-zip — fast, creates much smaller files, does both .tar and .gz, and, of course, free.
The usual utensils — computer, hard disk, mouse, keyboard, live internet connection.

Method

Download the tarball — the .tar.gz file from SixApart.
Unpack it on your local drive (use what you like, I just use 7-zip throughout). It will create a folder called MT-n.nn-ll where n.nn is the version number (like 3.35), and ll is the language code (en for English).
Rename this folder to mt (lower case), if mt is the name of the folder that contains your installation. If you’ve installed mt to the cgi-bin, then rename this folder to cgi-bin. This is important: remember to organize things so that you have a folder called mt and all the other folders and files of the installation below that.
Rename any files you’ve customized or edited in that the newly-renamed mt or cgi-bin folder that you’ve customized or edited. Specifically, these could be any standard files or, more likely, the default search templates. Rename them to anything you like, or they’ll over write your existing files.

Example: if you have a customized version of default.tmpl (the default search template in the search_templates folder), then, in the newly downloaded-unpacked-renamed folder (created as above), you need to rename default.tmpl to something like default-335-tmpl, so that it doesn’t over-write your custom version. Do this for all your customized files.
Zip it up. Pack up the new installation set that you have on your local drive (the one with the renamed folder and files) to create a .tar.gz file. I recommend you use the free utility 7-zip. If you are using 7-zip, see below for instructions on how to make this .tar.gz file.
Upload the newly baked .tar.gz file to a folder immediately above the mt installation on your server. If your mt is installed to /public_html/mt, then put the .tar.gz into /public_html.
Unpack on the server. From your CPanel, go to File Manager. Go into your public_html directory. Click on the uploaded .tar.gz file. In the top right corner, you will see an option for extracting the contents. Click that. Another window opens, showing the extraction from the .tar.gz. If you’ve followed instruction, your new files will go straight into your old mt installation, over-writing any existing files.

That’s it, you’re done! Now just go to your mt login page and the upgrade process begins.

Serve hot with a scoop of relief.

I find this a whole lot simpler than trying to FTP a 10MB installation, which is the size of the extracted mt installation. On an erratic connection, this can take hours. The other alternative, viz., uploading the factory-shipped .tar.gz to your server, extracting it, and then copying over individual files and folders is equally cumbersome.

The beauty of this is that this doesn’t delete any files. So if you’ve got other plugins, etc., already installed, they’ll all remain intact.

The topping. Creating a .tar.gz file with 7-zip

The reason I use 7-zip is that it gives me the option of doing the .tar.gz in one place. There are others that do this too (WinRAR didn’t, last I checked; you had to use it with a separate utility like WinGZ), but this one’s free and fast and creates significantly smaller archives.
First, right-click the mt folder (the one you’ve renamed). In the context menu, select 7-zip, and add to archive. In the 7-zip main interface, select .tar as the format.
It will generate a file called mt.tar.
Now right-click mt.tar and again select 7-zip and add to archive. This time select .gzip as your format option.
It generates the mt.tar.gz file.
Basically, you’ve zipped twice, once to .tar and then to .tar.gz. You’ll notice that the 10mb installation is now down to about 2.3 MB, much easier to deal with.

Hope this helps. Bon apetit!

A copy of this is on the SixApart community forum.

Migrating MovableType

2006-07-11T14:51:09Z

Recently, I shifted four MovableType blogs to my present host, Bluehost. These were four separate installations, on a personal unlimited license. The existing databases — the ones to be moved — were heavily populated: lots of plugins, customizations, plenty of data, including pictures and long blobs of text. I also had three or four users per blog.

The problem was this: how best to seamlessly migrate the blogs, retaining users, passwords and all customizations, without having to reconfigure MovableType from scratch?

After some trial-and-error, I figured out a way. I don’t claim that this is the best, or even the only, way in which to do this, but it worked for me and should work elsewhere, as long as the new host meets certain requirements (listed below). Here’s my recipe for shifting MovableType to a new host, based on what I did.EOB

]]> Moving a xMTx installation to another host: the recipe

Ingredients

First, some assumptions:

That you can access your MySQL database, and know how to use the export, import, drop, empty and SQL query functions. Very likely this is through PHPMyAdmin
That your target/new host runs the CPanel control panel. If you’re using anything else, I can’t help you, though the overall instructions may still be useful.
That you’re already running MT 3.2. This may not work for an upgrade; in fact, it very likely won’t, because, as far as I know, Version 3.2 uses a different MySQL table structure from the previous versions.

Separate the eggs: Backup and export your existing database.

This is critical. Log in to your existing MySQL and export the entire MT table. I prefer to do two exports, one each for the data and the table structure. But that’s only because it’s easier to upload at the other end. Remember to export with the create tables and inserts ticked on.

Heat the oven: Upload and install xMTx

On your new host, upload and install xMTx. For the upload, I strongly recommend using the Cpanel File Manager. It allows you to upload easily right from your browser, without going through an FTP client, and, more importantly, has a nifty built-in unpacking utility that will unzip/unpack your MT files. Remember also to upload any customizations and plugins you may have made.

What I do is this:

Upload the core MT distribution;
Unpack/unzip it;
Upload a tar.gz file of my customizations and plugins (which I generally keep in a separate directory anyway). Unpack them, and, if you’ve configured the tarball correctly, it will unpack right into the MT folder, over-writing where necessary.

Tip 1: If you’re as much a maniac as I am about using MT Plugins, then you’ll soon have realized that it’s damn near impossible keeping track of what goes where — the folder hierarchies, the customized files, etc. Besides, I also customize some of the core libraries so that my uploads go to pre-defined directories and all uploads come up with pre-designated code. If I muddle all this with a default MT distribution, I’m bound to lost track of some file essential to a plugin. So what I do is to keep a separate folder called mt on my local machine, and store all the plugins and customizations in that, preserving their required folder structures and placements. Then I create a tarball of this entire folder. Now, when I need to ‘customize’ a new MT installation, I have only to upload this folder and unpack it. Since the core distribution is always unpacked to a folder on the root called mt, unpacking the customizations-folder from the root (using Cpanel) will automatically add the plugins and customizations at their appropriate places.

Tip 2: to quickly create tarball files:

Use 7-zip (freeware) to create a .tar file;

Run 7-zip again on the .tar to turn it into a .tar.gz file (you can also use Win-GZ to turn the .tar into a .tar.gz file. Download Win-GZ.)

I’m not going to get into the details of the MT installation and initialization procedure, which are well set out in the MT documentation. Remember only that before you install xMTx, you must create the necessary database, database user, and add the user to the database with full privileges. The Cpanel interface is quite intuitive for that. Remember to update your mt-config.cgi as documented.

Blend at high speed: Enter your new xMTx installation

Okay, now that you’ve uploaded MT, your customizations and plugins and set all the correct permissions for the various .cgi files and such, initialize the system. After that, you’ll log in to the default MT interface using, of course, the default login (Melody) and password (Nelson).

Okay, you’re in. First off, create a new user, preferrably using the same login and password you had earlier. Remember to give this new user (you) full access to the default blog, ie, all privileges. Also remember to rename the default blog. Do one rebuild. I don’t think it’s strictly necessary, but it doesn’t harm. Don’t worry about paths and settings at this stage.

Now log off as Melody Nelson; log in again using the new user you created. Go to the authors page and delete Melody Nelson as a user.

Log off and close this browser window. Onward!

Add seasoning to taste: Upload additional files

I forgot to do this, so I’m making a note of it, though it’s self-evident. Rememebr to upload all the additional files for your website including:

Your css files (important!);
Your template files (important!);
downloadable files, if any;
image and picture files;
and, generally, anything that’s not stored in your database.

You can upload and unpack a zip file of these files, or use an FTP client (I use CuteFTP).

Pour into prepared pan: MySQL and PHPMyAdmin Adding the table structure

Here’s the next nifty bit. Enter your new host’s Cpanel again. Go to the MySQL page, and from there to the PHPMyAdmin page. You should be in your database. You may have to select your database if your host allows more than one, from the drop-down list. Select it.

Now select all the tables in the database and drop (read, delete) them. Open up a good text editor — I use TextPad — and open the MySQL table-structure file you exported from your previous installation. You can upload it directly into your MySQL database using the import function or, alternatively, use the SQL Query function and copy the entire table into it. Since your export has the create table and insert into lines in it, this will replicate your entire earlier table structure.

Do not repeat with the data export file just yet. You need to edit this file before you upload/insert the data (populate the table).

The critical thing now is to change the paths to your templates in your exported data file because it’s extremely unlikely that your earlier template-paths match the paths on your new host.

For example, earlier path name:

   /home/net/somehost/domain/html/templates

While the new path name might be

   /home/accountname/public_html/templates

Ideally, you should do a search-and-replace for the old path names and substitute them with the new path names in the exported data file.

Done? Fine. Now in PHPMyAdmin, upload or insert the contents of your freshly-edited exported-data file into MySQL.

The topping: full rebuild

That’s about it! Try logging in now, with the username and password you used on your previous host — you should be through without any difficulty. Remember to do one full rebuild.

Sit back and savour: tuck in!

All this sounds terribly complicated, but, believe me, it’s a whole lot simpler than trying to re-configure the xMTx installation from scratch.

Lastly, one important thing to note: do not try and merely shove the exported table/data into the database of the new host. You just won’t be able to access the xMTx interface, no matter what you do. You have to go through the process of installing and initializing the new xMTx installation, logging in, adding a user and so on, as described above. Frankly, I understand the first bit, but don’t quite know why it requires you to log in and add a user. But that’s a small thing, given the huge saving in time and effort. Doing it like this, I was able to move all four installations over to Bluehost in less than 30 minutes. Doing them over would have taken three days.

Character Entity References

2006-06-27T03:38:08Z

Over at A List Apart, one of the most delectable and beautifully presented websites for webdesigners, Peter Sheerin has a now legendary article titled “The Trouble With EM ’n EN (and Other Shady Characters)” on typography, typographical styles and character entity references. It’s a pity his advice isn’t followed more often throughout the Web. There’s little doubt that it makes for much easier on-screen reading. Apart from anything else, following his advice renders pages that are much more courteous to the reader.

One of the references in that article is to the W3C page on Character Entity References in HTML 4. There is a detailed table there of all character entity references with their HTML codes and CData numeric codes. Unfortunately, the table is difficult to decipher. I’ve taken that table and re-formatted it for easier reading. Presenting the W3C-based table of Character Entity References…

]]>

		html character entity	cdata numeric code	representation
		nbsp	160	no-break space = non-breaking space
¡	¡	iexcl	161	inverted exclamation mark
¢	¢	cent	162	cent sign
£	£	pound	163	pound sign
¤	¤	curren	164	currency sign
¥	¥	yen	165	yen sign = yuan sign
¦	¦	brvbar	166	broken bar = broken vertical bar
§	§	sect	167	section sign
¨	¨	uml	168	diaeresis = spacing diaeresis
©	©	copy	169	copyright sign
ª	ª	ordf	170	feminine ordinal indicator
«	«	laquo	171	left-pointing double angle quotation mark = left pointing guillemet
¬	¬	not	172	not sign
		shy	173	soft hyphen = discretionary hyphen
®	®	reg	174	registered sign = registered trade mark sign
¯	¯	macr	175	macron = spacing macron = overline = APL overbar
°	°	deg	176	degree sign
±	±	plusmn	177	plus-minus sign = plus-or-minus sign,U+00B1 ISOnum
²	²	sup2	178	superscript two = superscript digit two = squared
³	³	sup3	179	superscript three = superscript digit three = cubed
´	´	acute	180	acute accent = spacing acute
µ	µ	micro	181	micro sign
¶	¶	para	182	pilcrow sign = paragraph sign
·	·	middot	183	middle dot = Georgian comma = Greek middle dot
¸	¸	cedil	184	cedilla = spacing cedilla
¹	¹	sup1	185	superscript one = superscript digit one
º	º	ordm	186	masculine ordinal indicator
»	»	raquo	187	right-pointing double angle quotation mark = right pointing guillemet
¼	¼	frac14	188	vulgar fraction one quarter = fraction one quarter
½	½	frac12	189	vulgar fraction one half = fraction one half
¾	¾	frac34	190	vulgar fraction three quarters = fraction three quarters
¿	¿	iquest	191	inverted question mark = turned question mark
À	À	Agrave	192	latin capital letter A with grave = latin capital letter A grave
Á	Á	Aacute	193	latin capital letter A with acute
Â	Â	Acirc	194	latin capital letter A with circumflex
Ã	Ã	Atilde	195	latin capital letter A with tilde
Ä	Ä	Auml	196	latin capital letter A with diaeresis
Å	Å	Aring	197	latin capital letter A with ring above = latin capital letter A ring
Æ	Æ	AElig	198	latin capital letter AE = latin capital ligature AE
Ç	Ç	Ccedil	199	latin capital letter C with cedilla
È	È	Egrave	200	latin capital letter E with grave
É	É	Eacute	201	latin capital letter E with acute,U+00C9 ISOlat1
Ê	Ê	Ecirc	202	latin capital letter E with circumflex,U+00CA ISOlat1
Ë	Ë	Euml	203	latin capital letter E with diaeresis
Ì	Ì	Igrave	204	latin capital letter I with grave
Í	Í	Iacute	205	latin capital letter I with acute
Î	Î	Icirc	206	latin capital letter I with circumflex
Ï	Ï	Iuml	207	latin capital letter I with diaeresis
Ð	Ð	ETH	208	latin capital letter ETH
Ñ	Ñ	Ntilde	209	latin capital letter N with tilde
Ò	Ò	Ograve	210	latin capital letter O with grave
Ó	Ó	Oacute	211	latin capital letter O with acute
Ô	Ô	Ocirc	212	latin capital letter O with circumflex
Õ	Õ	Otilde	213	latin capital letter O with tilde
Ö	Ö	Ouml	214	latin capital letter O with diaeresis
×	×	times	215	multiplication sign
Ø	Ø	Oslash	216	latin capital letter O with stroke = latin capital letter O slash
Ù	Ù	Ugrave	217	latin capital letter U with grave
Ú	Ú	Uacute	218	latin capital letter U with acute
Û	Û	Ucirc	219	latin capital letter U with circumflex
Ü	Ü	Uuml	220	latin capital letter U with diaeresis
Ý	Ý	Yacute	221	latin capital letter Y with acute
Þ	Þ	THORN	222	latin capital letter THORN
ß	ß	szlig	223	latin small letter sharp s = ess-zed
à	à	agrave	224	latin small letter a with grave = latin small letter a grave
á	á	aacute	225	latin small letter a with acute,U+00E1 ISOlat1
â	â	acirc	226	latin small letter a with circumflex
ã	ã	atilde	227	latin small letter a with tilde
ä	ä	auml	228	latin small letter a with diaeresis
å	å	aring	229	latin small letter a with ring above= latin small letter a ring,U+00E5 ISOlat1
æ	æ	aelig	230	latin small letter ae = latin small ligature ae
ç	ç	ccedil	231	latin small letter c with cedilla
è	è	egrave	232	latin small letter e with grave
é	é	eacute	233	latin small letter e with acute
ê	ê	ecirc	234	latin small letter e with circumflex
ë	ë	euml	235	latin small letter e with diaeresis
ì	ì	igrave	236	latin small letter i with grave
í	í	iacute	237	latin small letter i with acute
î	î	icirc	238	latin small letter i with circumflex
ï	ï	iuml	239	latin small letter i with diaeresis
ð	ð	eth	240	latin small letter eth
ñ	ñ	ntilde	241	latin small letter n with tilde
ò	ò	ograve	242	latin small letter o with grave
ó	ó	oacute	243	latin small letter o with acute
ô	ô	ocirc	244	latin small letter o with circumflex
õ	õ	otilde	245	latin small letter o with tilde
ö	ö	ouml	246	latin small letter o with diaeresis
÷	÷	divide	247	division sign
ø	ø	oslash	248	latin small letter o with stroke
ù	ù	ugrave	249	latin small letter u with grave
ú	ú	uacute	250	latin small letter u with acute
û	û	ucirc	251	latin small letter u with circumflex
ü	ü	uuml	252	latin small letter u with diaeresis
ý	ý	yacute	253	latin small letter y with acute
þ	þ	thorn	254	latin small letter thorn
ÿ	ÿ	yuml	255	latin small letter y with diaeresis

Notes:

For HTML Entities, use ampersand & and semicolon ; before and after the given code. Thus, nbsp should be written
For numeric data, precede the code with ampersand & and hash/pound #, and follow it with a semi-colon ; Example: ù

I hope you find this useful!

Movable Type - Related Entries

2006-01-30T12:29:27Z

I’ve long been less than fully happy with the ordinary Related Entries plugin for MovableType. Originally developed by Adam Kalsey, it is now being maintained by Tim Appnel. Tim’s instruction was on the mark — running MT-Related with keywords slows rebuild times and increases overhead enormously.

Back to Adam. Found he had a brilliant (hardly surprising) alternative using Brad Choate’s MTSQL plugin (are these guys gods or what?)

]]> Naturally, I ran into trouble. Not being a SQL-ey kind of guy, I pottered around in this briar patch for a bit, and only came away bloodied. Of course I had to decide to crank it up a notch and go for the fancy stuff — how to suppress the related entries block if there are none; how to set a threshold limit, etc. etc.

Now Adam has a bit on his further comments section on all this, but, somehow, none of it really worked for me. In particular, I couldn’t get his use of Brad’s MTIfEmpty plugin to work at all.

I found Adam’s recipe for suppressing display using MTIfNotEmpty, and for using the score > 5 didn’t work well for me. I kept getting one SQL error after the other. I won’t bore with you with the details, but here it is as I see it (remember, I am not a sql-ey!):

AND SCORE > 5 will never work. It has to be HAVING SCORE > 5

That’s one thing. The second thing is that using the tag inside the tag invariably returned an error. I had a hellish time trying to escape all the single and double quotes, the various types of brackets and braces, and really got nowhere.

What then?

Sometimes, as they say, the simplest answer is the best. Here’s my workaround. It works for me, and should, logically, work for anyone else. I’m not a programmer or a designer, so don’t ask me why — I just did what I thought was logical!

Related Entries <$MTEntryTitle$>

What happens is really a two-step check. First it checks if you have any keywords or not; if you don’t, nothing displays. If you do, then you have to pass the second threshold, of a score greater than (or greater than or equal to, in the above example) a figure that you determine. Once you do this, it works fine.

Note that the correct usage seems to be HAVING SCORE > whatever (you will have to use >). And, of course, you have to get rid of the silly div tags which are from my own little CSS garden!

Multiple Blogs on MT

2003-12-02T04:36:29Z

Heaven! Finally managed to get a couple of php scripts working to my satisfaction for my blog suite; suite being the operative word. Of course, it’s powered by Movable Type (you mean there’s anything else? … more on that anon) and while I really love MT (reasons, too, anon), what I really wanted was much more organisation and hierarchical structuring than just a single blog.

The categories feature didn’t really do it for me; not enough anyway. You still had a single page with all kinds of stuff on it and I’ve found hunting through blogs can be really tedious.

]]> What I thought was this: I’d have a set of blogs. Each blog covers one major area of interest: thus, politics, history, art, books/language, humour, food & wine, and so on. The posts in each blog are categorised. The geekspeak blog, for example, deals with technology and the categories include email, blogging etc. This is kind of neat and tidy — you don’t end up looking in posts on history and Dubya for stuff on blogging and PHP.

THE COLLATION PROBLEM

The real trick was, however, to default the first main blog (blog_id= 1) into a non-blog blog. That is, it’s a blog for the purposes of configuration in MT, but that’s all. It has a single page (the main index file) no archives and no entries are made to this blog. Instead, this blog pulls material from the other blogs. In short, I have a multi-blog MT installation, with one parent blog and the rest as ‘feeder’ blogs from which the parent blogs pulls its material.

THE MULTIPLE BLOG SCENARIO

So far so good; but, immediately, trouble. How do you get the parent blog to constantly refresh its material? What I wanted was this:

a way of retaining the formatting on the parent blog page;
some way of pulling entries from the other blogs;
a method for fetching specific, important data (eg, the permalinks) for the entries from the other blogs;
some means of specifiying the maximum number of entries to exclude;
some way of doing all this automatically without having to rebuild the parent blog;

THE EXISTING SOLUTIONS IN MOVABLE TYPE

MT does have some solutions in the form of MT plugins , notably from David Raynes. His Other Blog has some of these features.

Unfortunately, it has two significant limitations: first, it demands that the blog on which it is used (the target blog, in this case my home page ) must be rebuilt to correctly reflect the updates on the source blogs. Second, it completely cracks up when dealing with any sort of archive. It just won’t work outside an index template.

That’s not good. So, I dug deeper and found two more plugins: David Raynes’ own Whole System and Stepan Riha’s Global Listings plugin. I didn’t try either because I figured that almost anything that is an MT plugin must demand a rebuild; and that I most certainly don’t want.

Both are said to be similar to a straight PHP script (not an MT Plugin) from the amazing Scripty Goddess. She has an absolute cracker of a MT hack, called the Multiple Blog Updates List. Her script page also references two other PHP solutions for ordered updates and a portal tutorial, both worth checking out. I looked at these briefly and didn’t like them. The ordered updates thread uses Server Side Includes and shtml files, which I wanted to avoid since they don’t directly reference the mysql database; and the portal tutorial also seems to rely quite a lot on MT Tags (maybe I’m wrong, but that’s the impression I got).

TWEAKING THE SCRIPTY GODDESS

:-) That’s in a lighter vein, I hope she won’t mind the caption. Anyway, I took the SG script and it worked all right, but I realized soon enough that it needed major tweaking. In its original form, it has at least these limitations:

It doesn’t allow you to specify a maximum number of entries on the main page; in fact it takes only the latest from each. I found I wanted more;
It has a problem with empty blogs;
It doesn’t correctly reference the permalinks for individual entries;

I rewrote large chunks of the script, and now I use this:

//connection info include ("/users/mcavity.com/httpdocs/connect.php"); //for each of the selected blogs (2, 3, 4) in mt_blog, find and display in descending order the most recent post in mt_entry. $alldata = array(); $alldates = array(); $blogcount=0; // keep track of number of blogs $blogarray = "SELECT blog_id, blog_site_url, blog_name, blog_description FROM mt_blog WHERE NOT (blog_id = 1)"; $resultblog = mysql_query($blogarray) or die (mysql_error()); while ($rowblog = mysql_fetch_array($resultblog)) { $burl = ($rowblog['blog_site_url']); $bname = ($rowblog['blog_name']); $bdesc = ($rowblog['blog_description']); $rowarray = "SELECT entry_created_on, entry_id, entry_blog_id, entry_title FROM mt_entry WHERE (entry_blog_id = (".$rowblog['blog_id'].")) ORDER BY entry_created_on DESC"; $resultid = mysql_query($rowarray) or die (mysql_error()); if (mysql_num_rows($resultid) > 0) { $count=0; while ($rowid = mysql_fetch_array($resultid)) { // if ($count>0) { break; } $maxarray = "SELECT entry_title, entry_blog_id, entry_id, entry_created_on, entry_excerpt, entry_text, entry_text_more, trackback_blog_id, trackback_created_on, trackback_entry_id, trackback_url FROM mt_entry, mt_trackback WHERE entry_id = ".$rowid['entry_id']." and trackback_entry_id = entry_id ORDER by trackback_created_on DESC"; $resultmax = mysql_query($maxarray) or die (mysql_error()); while ($rowmax = mysql_fetch_array($resultmax)) { $ids = array($rowmax['entry_id']); $blogid = ($rowmax['entry_blog_id']); $tburl = ($rowmax['trackback_url']); $date = ($rowmax['entry_created_on']); $title = ($rowmax['entry_title']); $excerpt = ($rowmax['entry_excerpt']); $text = ($rowmax['entry_text']); $textmore = ($rowmax['entry_text_more']); //saves the data int alldata array for use outside of the while loop $alldates[$blogcount++] = $date; $alldata[$blogcount++] = array('bname'=>$bname, 'blogid'=>$blogid, 'burl'=>$burl, 'bdesc'=>$bdesc, 'ids'=>$ids, 'title'=> $title, 'excerpt'=>$excerpt, 'text'=>$text, 'textmore'=>$textmore, 'tburl'=>$tburl); } $count++; } } } array_multisort($alldates, SORT_DESC, $alldata); for ($i=0;$i<=5; $i++) { //separate date created on up between year, month, and date $year[$i] = substr($alldates[$i], 0, 4); $month[$i] = substr($alldates[$i], 5, 2); $day[$i] = substr($alldates[$i], 8, 2); if ( $alldata[$i]['text']>""): echo '', date( " F j, Y", strtotime($alldates[$i])), ' : ', 'In '; echo '', $alldata[$i]['bname'], ' '; echo '', '',$alldata[$i]['title'], ' '; echo '', $alldata[$i] ['excerpt'], ' '; if ( $alldata[$i]['textmore']>""): echo '
', $alldata[$i]['text'], '', ' [ More ] ', '', '< /div>'; else: echo '', $alldata[$i]['text'], ' '; endif; echo '',' ', ' '; echo '',' ', ' '; else: break; endif; } ?>

Basically, this does several nifty things.

First, it includes all blogs except the first parent blog (the “WHERE NOT blog_id=1” line).
Second, it correctly identifies the permalinks. It first fetches all the blog_ids. Then it fetches the entry data for each blog by matching the entry_blog_id to the blog_id. Then it matches the entry_id in the next array (“maxarray”) to the entry_id in the previous array (which matched the blog_ids) and also requires a match of the trackback_entry_id to the entry_id — and that’s what yields the correct permalink.
It gets rid of the trouble caused by the “if ($count>0) { break; } ” line;
It allows you to set a maximum number of entries (setting the variable $i to <= 5 is the key; 5 can be any number you like, but be warned that because of <= and $i++, it will generate one more entry than the actual maximum);
And here’s the niftiest bit: if there is actually a text body, then and only then does it print it. This is necessary because on account of all that cross-matching above, it is otherwise apt to produce a whole series of blank entries while it loops.

I added a little refinement in the printing section: if there is an extended body, it insets [more] with a permalink http reference tag; and if there isn’t a an extended body, it just prints the main text entry.

I’m sure it can be improved and made more efficient by someone who knows PHP well (I’m really a newbie at this, just got into it a few weeks ago and I’m not even a computer jock, just an ornery lawyer) but it works well for me. If you’re using it, remember to remove all the class="whatever" tags and add your own — these only reference my CSS file entries. Incidentally, you can embed MT tags into the print section, too, but what that does is yet another saga.

THE SIDEBAR REARS ITS HEAD

So far so good. But there endeth the saga not. I also want a method of pulling the blog names, with their URLs, and the top five entries in each blog and sticking these into a little side-bar for navigation. I wanted this to serve the purpose of a navigation bar or menu (you add a blog, it must show up) and must, again, refresh itself.

MTOther blog failed totally because of the demand for a rebuild. I tried other solutions and got some really bizarre results. Then, on the footing that the simplest solution is often the best, I just used a PHP list command coupled with a while command:

//connection info include ("/users/mcavity.com/httpdocs/connect.php"); $result = mysql_query("SELECT blog_id, blog_site_url, blog_name, blog_description FROM mt_blog WHERE (blog_id > 1) ORDER BY blog_name ASC"); while (list($blog_id, $blog_site_url, $blog_name, $blog_description) = mysql_fetch_row($result)) { echo '+ ', '', $blog_name, ' '; $entryresult=mysql_query("SELECT entry_title, entry_blog_id, entry_id, entry_excerpt, trackback_entry_id, trackback_url FROM mt_entry, mt_trackback WHERE ((entry_blog_id = $blog_id) AND (entry_id = trackback_entry_id)) ORDER BY entry_title ASC"); while (list($entry_title, $entry_blog_id, $entry_id, $entry_excerpt, $trackback_entry_id, $trackback_url) = mysql_fetch_row($entryresult)) { if (mysql_num_rows($entryresult) < 6): echo '? ', '', $entry_title, ' '; else: break; endif; } } ?>

What this does is that it first gets all the blog names and lists them alphabetically. So, you add a new blog, it gets added at once. Next, immediately after listing the blog (the WHILE statement) it gets a list of entries where the entry blog id matches the blog id (which it just printed) and where the entry id has a matching trackback id (to get the permalink). Then it prints the sequence of those entries subject to a cap of 5 (the ending if statement where the row count of the entryresult is < 6): your last 5 entries.

THE REAL TRICK

And here’s the trick. Don’t use the MTInclude tag for these php chunks — again that forces rebuilds. Instead use the simple php include statement so that it automatically builds these list with every page load.

Happy, happy day! Perfect. Time to go grab that big bowl of cappucino. Yes! Where am I using this? On this site, of course, but also, with a variant, on the Bombay Bar Association website.

Postscript: Having done all that, I knocked it off. Frankly, I went back to a single blog system — I like the bookmarklet feature and, for a personal blog, it just wasn’t worth the effort struggling with multiple blogs.

Post-Postscript and Caveat: This all obviously horribly dated. With MT4 and MT5 multiblog is built right in, and MT5 is, in fact, designed with exactly this in mind. But it’s good to know I was aiming in the right direction — seven years ago!