Aeracode: ByteHoard

Thumbnails

I've literally just finished implementing thumbnail support in ByteHoard/e. Unlike previous versions, they're not generated on the fly, but rather whenever the file changes - the same goes for all the file metadata ByteHoard uses, like mimetypes (which now work, too) and file sizes. As a supplement to the thumbnails, two other features have appeared. Firstly, files now have relevant icons and type descriptions, and secondly, the administrator interface has a button to entirely rescan all the files a user owns, for those cases where you've managed to confuse ByteHoard. I've also closed a few bugs, including quota usage not changing when you delete directories, and not being able to remove administrator abilities from users.

The Aftermath

Well, the ByteHoard/e first alpha seems to have gone quite well. There were, as I expected, a few bugs - thanks to the people who found them, and I encourage anyone who finds anything slightly off to go and make full use of the new bugtracker at bugs.bytehoard.org. I'm now working on the second alpha, and I'll be releasing it... when it's ready. Expect filemail, registration, thumbnails, search, and more in this new release.

Alpha Zero

So, the first ByteHoard/e alpha is out, and is the result of quite a bit of hard graft. There are still some compatability issues, as there are with all scripts this complex, but I think it's a good first try. Nearly all the 'good' features are missing; search, registration, cross-user access, translation (although this is more a case of writing the script to generate the .pot files), and so on. However, I'll be working on these next. For those of you who actually read these blog entries, I'd love to hear any feedback about BH/e you have. Too complicated? Too simple? Not compatible enough? A step in the right direction? It's nice to hear responses, since you people probably have a better idea of what you want than I do. As I said above, translations will start shortly. This time round I'm using GNU gettext format, so I have to bodge together a script that will do both the PHP and the JavaScript collections simultaneously, and then split them up again. Well, onwards and forwards.

The Home Straight

The alpha release of ByteHoard/e is oh so nearly ready; all that is left is pretty much one function, the "add user" function, and possibly registration if I think it's necessary (it probably is). Recent changes have included the whole admin interface being added (which is also all AJAX code; I have a feeling I may regret this, but we'll see), the addition of a user preferences page, which so far only lets you change your password, full name and so on, and a new themeing and UI redesign so that the whole thing is a lot easier to look at and use. Screenshots of the latest version are below.

Making Real Progress

More work on the uploader today, and with some experience gleaned from the last two days' worth of fiddling about with uploads I now have a lovely asynchronous uploading panel. You can upload multiple files at once. You can see how far each individual file has got (i.e. a progress indicator; this does require a Perl CGI script, but there's no real way around that). If files fail, they turn red and you get a reason why (normally you've run out of quota). If files succeed, they turn green and then clear themselves after a few seconds (the disappearing is customisable for both fail and success). If you don't have CGI support on your webserver, it falls back to normal PHP uploading (which is actually slightly faster), and a little spinner animation. Overall, it's pretty neat, and is as close to Web 2.0 as ByteHoard uploading is probably going to get. Of course, behind the scenes it uses IFrames, but then that's the only way really. Of course, there is still the issue of uploading entire folders and drag-and-drop from your favourite file manager, but they are issues that can only be solved in some sort of client-side program, be it a java applet or a small executable file. At least I've finally got progress bars after three years. Screenshots follow, as usual.

Uploads & Downloads

So, after a touch more work on ByteHoard3 (well, perhaps a bit more) I have uploads and downloads working, along with deletion. The usual screenshots are attached, but basically the uploader currently creates little progress bars below the upload form with details about each file, which turn green or red at the end depending on completion or failure. The download works exactly as all other downloads do - click and your browser bugs you about what you should do, and optionally how evil files might be. The delete action has an inline confirmation, although I may have to redesign it as I develop on a widescreen and it doesn't like anything slimmer at the moment.

Miles of Files

I promise I'll stop thinking of terrible names for blog posts at some point. Time for a ByteHoard update. I've given it a bit of attention today, and so, among other things, we now have:

Folder/file navigation working completely; the folder tree, breadcrumb bar and subfolder list all work (this will make a lot more sense with a screenshot; see below)
Sorting (a feature sorely missing for the last few years) makes a strong entrance. It currently only sorts by name or size (ascending or descending), but I'll throw in types and anything else I can think of later.
The interface has been tweaked so everything lines up (finally, a justified use of tables) and even the logout works.

There is still much work to be done, but ByteHoard is currently a very nice, fast and useable file lister. I'll be working on adding upload and download next. As a special reward for reading this far, you get rewarded with a screenshot of the current files interface. This may change, but probably not too much, at least visually.

Folders Grow On Trees

I've just finished the folder tree panel for BH/e, and because I'm a kind, caring developer (ahem), you get some pictures... It's your basic AJAXy tree system, where you click on a folder's > picture, a little loading spinner appears, and then shortly afterwards the subdirectories appear below. Currently it reloads every time you collapse and re-expand a folder; I'll be working on making that more efficient later. I've been debating about where to put the folder tree view and if I should integrate it with the main file view and have everything all in one. I think it's best to have it as a sidebar, both because people are used to that, and because it means having a lot of files in a directory doesn't push the next lot of folders off the bottom of the screen. I might stick a breadcrumb bar in above the file list, though.

Filesystem Engineering

More leaps and bounds (or, to give them a better name, tiny footsteps) were made with BH/e today. The filesystem 'engine' is pretty much done now, and it's a lot better than the BH2 one. It's your standard files/folders heirarchy, but with a windows-esque separation into several top-level directories (i.e. 'Your Files', 'Andrew's Files'). There's none of the database buildup caused by bad implementation before, either, and as a plus there's folder sizes (as in actual, recursive ones) with very little processing power thanks to a gradual update system. There's even a rescan function if it all goes horribly wrong, which works quite well as well. Now I need to start on the interface to this lot.

Re-Re-Redesign

I've redesigned the ByteHoard website again; for those of you who've missed its last two years this is about the seventh design. This new one focuses on clean lines and readability, as well as saving a lot of bandwidth. CSS is still a bit behind where it should be, although that's obviously not the specifications' fault (I'm looking at you, IE). However, I only had to make one change to my CSS to get IE to play nice (and it's not a hack, I just had to set only one of a bunch of nested elements to display:block), so there is that. Apart from that it seems to have gone well. My usual basic txt CMS is driving it, with my new submenus which I first developed for the Aeracode main site thrown in too. With some RSS fetching optimisation (i.e. only do it on the news page) it's also a lot faster. The logo also got an overhaul; I finally found the right font, NeoSans, and used that to make a much cleaner logo. Overall, it went pretty well, I think (well, for only 4 hours of coding that is). Now back to ByteHoard proper.