Update
Argh, it’s the early hours of Wednesday morning and the column was meant to be updated on Monday. Weird thing is I already have a draft of about 70% of it. Oh well, I’ll finish it later today (Wednesday) and post it, back-dating it to Monday and I’ll let it run for 2 weeks.
Meantime, things have seemed a little slow because I’ve been concentrating most of the last 2 days or so on the admin system design work. Now, seems a little excessive perhaps, but it’s not.
Let me explain a little more of what’s involved in the admin system and why I’m spending the time on getting the structure right in my head before I even touch a line of code.
The admin system is designed with the express purpose of being able to update and add strips via a web interface. That means, first up, moving the code that I already relocated back to my PC… back to the server.
That also means resetting up a database for WordOwl to use, but this time instead of just having raw data to contend with, it also has to manage user logins, permissions, archival of strip data, and somehow integrate that into Sphinx, as well as cache data to avoid the issues I originally had.
If that wasn’t enough, the other problem is the transcribing interface; since mid February last year, I’ve been using a customised OpenOffice spreadsheet to build XML. It’s functional, it works reasonably well. But now I’m using a web interface, the choice of OOo is no longer a valid one.
So after a long conversation with Sophie, a strong comic transcription format has been established. No, it isn’t Oh No Robot’s format, it’s one that much better suits what WordOwl needs. There’s also a simpler version going to be available too, which is pretty neat.
It’s late, and I’m not sure I’m making one of the biggest points; it’s not an update I’m doing, it’s not a new set of features surgically implemented on top of the existing code. It’s likely to end up being a serious rewrite. As it stands, the entire back-end will be rewritten and integrated into Sphinx that much better, while the front-end will need serious patches to suit the new format.
I am also ditching some things that just aren’t needed.
To go out:
- A distinction between ‘inset’ and ‘caption’ text; the distinction is so subtle it’s not worth worrying out, so it’s going.
- A separate ‘keywords’ option on the search panel.
- The AJAX-like dynamic loader interface to the search page. Not only is it causing code duplication, the fact that it effectively breaks the ‘back button’ and that there isn’t a suitably clean solution to it means it can go. It’ll actually save some bandwidth overall too, but that’s a side issue.
To go in:
- The ability to create, edit and delete strips as listed in the database and have those changes applied very, very quickly.
- The ability for comic authors to designate transcribers for a comic and for them to edit the transcription before going live; any changes must be approved by an appropriate user and a non-admin user will not be able to approve their own changes.
- Ability to search by author as well [NB despite the number of hours on the design documents I currently have, I have not yet specified the details on how this will actually work, nor how it will look in the interface, however it is the intention for it to function as any other search option]
The thing about the admin system is not only the complexity of it, but for the first time I’m actually going to be using a proper MVC framework - CodeIgniter. Note that this won’t be rolled out to the entire site, simply the admin area; the rest of the site can rely on my more custom codebase since speed is more important there than it is on the admin area, as it is being used by many more people.
Having read the manual for CodeIgniter, it should reduce a lot of the grunt work in building what is essentially a straightforward web app. Ultimately we’re talking displaying the contents of DB tables on a form and editing them. Nothing weird there, actually, the only real weirdness is in the logic area in maintaining the state tables, since I’m having multiple tables that cache data in different ways.
I do also need to sit down and design the interface for the admin section; I’m still not even remotely sure how that’s going to look yet, though I am thinking perhaps some kind of dashboard.
Although the AJAX-interface is gone from the search page, the admin area is very likely to use more of it for the interface; it is much more suited to needing it, really.
I have to be honest, I’ve had so many thoughts about how to make this part operate it’s unreal. Being honest? I even thought about buying my own (cheap) SSL certificate for the admin area. Don’t really know why, just seemed like a good idea, though I haven’t done so because as much of a good idea it may be it’s just not practical.
I also promised in my last update a little news on Angst Tech. Well, due to hosting issues, Barry of InkTank is currently migrating all of the InkTank comics over to a new host, with a new codebase. As a result, every URL will move and need to be rewritten, so what will happen once the move is announced is that I will remove every InkTank strip from WordOwl, until I can reintroduce them unilaterally (i.e. I’ll strip out all of InkTank, Sorry We’re Open, Weak-End Warriors and Angst Tech and as I fix the URLs, I’ll re-add them)
Anyway, it’s nearly 3am and I need to go to bed.