IMS_Blog

Browsing the archive can produce a page on a particular topic, with a list of available photographs and thumbnails of that image. As I browsed I noticed something: Those thumbnails were taking quite a while to load. And, actually, Firefox seemed to be using quite a lot of memory, as if it had loaded a lot of large images, but I wasn’t looking at individual pictures, just the list….as if loading a lot of large images.

Oh yes, those ahead of me will already be there. Those aren’t thumbnails in the list. Each page is loading the complete, full size image, all they’ve done is set a height and width attribute on the img tag so the browser resizes it. For fuck’s sake, this is really web 101. If you want to serve a thumbnail of an image, you need an actual thumbnail to serve. Serving the large images and simply allowing the browser to resize client side has several, very well known issues. First, as I noticed, it simply slows everything down horribly, particularly if the user doesn’t have a blisteringly fast connection (any dial up users you may have now can’t even choose which images they’re going to sit and wait an age to load, because they have to wait a much longer age for them all to load in the first place). Second, it hurts in terms of bandwidth. This may be a problem for those with a limited monthly limit (maybe not, they’re only images, but there are a lot of them) but it is certainly a problem for pay as you go dial up users, or even pay per bit broadband users if there is such a thing (I think at least some ISPs will allow you over that monthly cap but then charge for the amount of data downloaded). It also hurts your own bandwidth, as you constantly serve huge images up to those who just want a tiny slick version. It’s not even difficult. There are various pieces of software to handle everything to automagically resize images to thumbnail proportions when requested (incidently, the individual pages do exactly the same trick to get images to a standard size for the page. Strangely though there doesn’t seem to be an actual link to the full size image, but it’s easy to see that’s really what’s going on by doing a right click-View Image, or over-riding the height/width attribute etc).

There are some other peculiarities too. Looking at a category in list view (erm, such as this one) produces some meaningless column labels (what exactly is “Term1 1” or “Thes system”?—well, actually I assume Thes system has something to do with the fact they keep referring to the system as a “Thesauri”—that page has a side bar item labelled “Popular Thesauri”. No, I don’t know quite why. What was wrong with “Popular Categories”?) It’s not at all obvious what the difference between list, image list, and detailed views should be either. Clicking detailed seems to get to the same place as clicking view, though with slightly different url parameters so maybe I’m missing something. All those topics seem to be in the system as “subject headings”, yet they still insist on telling the user constantly that it’s a subject heading. That would be fair enough, were it ever anything else.

Then there is the 20 item limit. I’ve never seen a subject with more than 20 items “directly related”. I have seen a lot with exactly 20 items though, which is somewhat suspicious, as if there ought to be a second page or something. And then consider the subject heading streets. It has exactly 20 directly related items, but notice the list underneath, a whole load of somethings all called Streets and all (well, my random sampling) containing precisely nothing. I bet that’s where all those other pictures after number twenty are supposed to be hiding.

Oh, the server occasionally flakes out with maximum execution time errors too. Maybe that’s why the registration to use it’s promised “advanced features” doesn’t work, it’s claims to have emailed me nothing but broken promises.

In short the entire thing feels like a beta (or maybe even alpha) in the old sense, rather than the spanking new GMail etc sense. One has to wonder if that site has undergone any user testing, any controlled release to, say, access from the library only to test it out. I want to give them the benefit of the doubt and assume that they are aware it’s not perfect but felt it better to give the public some access to a half working site than no access at all. Maybe.

I haven’t even mentioned the underlying code, with it’s mix of “use divs and css” because it’s modern and old school table layout, without any sort of semantics. Or the fact that while it proudly carries an html compliant button the developer hasn’t actually gone that finally yard to get rid of the few remaining errors from the validator (all of which seem to be thanks to stray unescaped characters, & instead of & etc).

I’ve just sent a (slightly more polite than this post) email mentioning the worst problems (stupidly large images for thumbnails), so we’ll see if there’s any response (there isn’t an obvious contact address on that whole site, so it got sent to the generic libraries@gateshead.gov.uk with a request to forward it on to whoever is in charge). I have to confess I did pull qualifications and sign off as Dr in the hope it might just get more attention though ;-)

Aside: Not directly relevant but the whole thing claims to be powered by something called Trio. I had a look at the software’s website. All the supplemental pages (pricing info etc) are currently just throwing ASP error pages. Hardly inspires confidence…

Update: Well, at least somebody in the library takes the time to reply. Apparently my comments have been forwarded to the Reference and Information Manager. Allegedly someone will contact me in due course, probably to say thank you for your comments again but at least they made the effort to reply.

Update 2: I have to confess to being quietly impressed with at least the speed of the response, and even some of its content. Another email back from the people in charge (the Project Officer no less). It claims they’re waiting for a patch from the website suppliers to fix various problems, like the subject headings issue, and they have low resolution (proper) thumbnails they intended to put up with said patch. At least it goes on to say they may have to consider getting those up sooner. Well, yes. Frankly the site should never have been allowed to go live without them. The other technical problems you can almost forgive but to so quickly have to patch does suggest a severe lack of user testing, or an ill judged decision to release a half baked product. Perhaps getting that patch isn’t helped by the Trio technical support pages still being broken themselves…Rest assured dear reader that we shall be monitoring progress with interest.

Update 3: It can’t be said they don’t take comments seriously. Another reply, from a different person. It basically says the same things but at least everyone seems to be in the loop and know what’s going on. However I must say I still don’t believe that the site should have been allowed to go live for public access in it’s current state. At least it sounds like it’s getting sorted…