I just started work on a behaviour for a new workflow in Fez and had a quick fish around for another workflow to base it on. I realised that for a ‘non-auto’ behaviour, preview.php is a good place to look as it just has the workflow wrapper code and then includes another file.

 include_once("../config.inc.php");
 include_once(APP_INC_PATH . "class.template.php");
 include_once(APP_INC_PATH . "class.workflow_status.php");
 // include whatever else you need here
 $tpl = new Template_API();
 // set the workflow wrapper template
 $tpl->setTemplate("workflow/index.tpl.html");  
 // set the template that the workflow template will include
 $tpl->assign('type',"preview"); 
 // restore WorkflowStatus object from the session
 $wfstatus = &WorkflowStatusStatic::getSession(); 
 $pid = $wfstatus->pid;
 // set up the basic templates variable you need for a workflow
 $wfstatus->setTemplateVars($tpl);  
 // handle any form input processing here
 // check if the user clicked a button to go to another state
 $wfstatus->checkStateChange();  
 // Do all your display work for generating the form and page content
 $tpl->displayTemplateRecord($pid);

Simple eh?

Do you have word docs in your Fez install? Want users to be able to fulltext search em? Well Jonathan Harker of Catalyst recently added support for Word doc fulltext indexing and it wasn’t that hard! I’ve added the details to the Fez Wiki:

Fulltext Index Word Docs – FezWiki

First you need a way to convert the document to plain text. This example uses Antiword but it is also possible with catdoc.

Are you a true Fez fanatic? Well that must make two of us! In case the fez-users and fez-developers mailing lists aren’t enough for you. You can also get the following RSS feeds to keep up with it all:

• Fez code subversion repository for any code we commit, including comments, time and who did it: Fez SVN RSS
• Fez Wiki changes
• UQ eSpace news
• Oh yes, and my blog RSS right here

Guess what. You can write some really nifty repository features when you’ve got about a hundred or so test objects in your development server. Sometimes like we did earlier this week, you might role ‘em out live and get some unpredictable results. Despite testing on a staging server first, UQ eSpace took some heavy hits this week once three search engine indexers and a whole bunch of users got stuck into the changes that went live on Monday afternoon. Three days later (with a public holiday in the middle) and we are just getting to the end of some of the nimble keyboard action that we needed to do to get things under control again.

Of course we could have rolled back to an earlier version but we felt that we could get things up to speed again fairly quickly and would rather do that than have to do several back and forth upgrade / downgrades while we tried to get it right and let all our other project timelines slip away away away.

The good news is that we now have a more battle hardened UQ eSpace with some nicer features. 8956 records and growing everyday!

The Fez repository software has been very successful and generated a lot of interest from groups internal to UQ as well as other universities. The result is that even with our new programmer, we have so much work to do and so little time that we don’t always get around to helping those out there in fez-users mailing list land.

I spent today reviewing and writing requirements for additional stuff we need to do for an internal customer and I’ve ended up splitting it into three projects! So we may be getting even more programmers.

One hopes anyway.

There has been a part of the Fez workflow that has been bugging me. The original page I made to select a record as part of a workflow used three drop down boxes with ajax behind them to select first the community, then the collection and then the record. The problem was that sometimes there are more than 300 items in a collection so the drop down boxes required a lot of scrolling if the ajax didn’t just timeout.

Today I made it so that you can type the title of the record you are after and it will populate a short list of candidates from which the user can select the record they want. I originally made it so that it showed a list of the record citations but this proved too slow. So it just lists the PIDs and titles of records that match the typed simple search terms.

I’ve just been doing a bit of work on an icon for Fez object cloning. I picked an icon which shows two pieces of paper next to each other but on the white background they were hard to see. I decided to add a grey tint the icon so it would be more visible.

As a bit of an open source junky, I thought I’d have a go at Paint.net as a lightweight image editor. I opened the icon, created a new layer, selected the two bits of paper with the wand tool, pasted grey onto the top layer and then changed the transparency of the layer until I could see the paper icon underneath. Paint.net was pretty easy to use and it’s free so I recommend if you are just doing simple tasks and don’t want to pay a license for full-on image editing software.

Now we can start on the Fez 1.4 release…

Wiki entry

Fez 1.3 is getting very close now. We’ve decided not to put out a Fez 1.3 RC4 so there’ll only be bug fixes for what we’ve found in Fez 1.3 RC1-3 and the testing that Lachlan and I are doing in Windows right now.

The hold up at the moment is that mime_content_type on windows just doesn’t want to work. Which means that files can’t be ingested (because a blank MIME type upsets fedora). We used to use the browser to get the MIME type but it is not reliable. I’m trying to come up with the best hack to fix it without the need for more PHP extensions or external applications. The best bet at the moment is to use JHOVE when ingesting to get the MIME type. Usually we run JHOVE after the fact but now we’ll shift towards using it to get some information about the file we’ll be ingesting. However, JHOVE is limited in what file types it knows so I’m trying to see if there’s a combo workaround using a few methods to come up with a MIME type. Maybe I can have some optional extras – like if you have the win32 file command or something.

Either way, we’re aiming for Fez 1.3 release on Monday 12th.

Releasing Fez 1.3 has been a difficult process because we can always keep finding and fixing more and more bugs and while we’re focusing on that, there are heaps of new features that people are looking for too. Many bugs that are coming up now are just annoyances that are not going to affect people too much where as others require big architecture changes such as the problem of having two images like basename.jpg and basename.gif getting jumbled in the derived datastreams (as fez will generate a thumbnail_basename.jpg that is shared by both images and depends on the order in which they are processed as to which image gets a thumbnail). However, I don’t think this problems has actually affected anyone.

Then there are a few things that we only wrote a basic skeleton for and now need to rewrite such as the searching.

We have acquired a new programmer, Lachlan who started last week and is finding his way into the system by getting it running on windows.

So i think we will be releasing Fez 1.3 RC4 later in the week and it will very likely become Fez 1.3 final.

In the meantime, more sites have been cropping up around the place where Fez is being trialled. The latest is MatDL . It’s always interesting to get an email out of the blue on the fez-users list (or sometimes direct) with a signature from a university or organisation whom we had no idea were using Fez.