The Corpus Callosum

The Utility I Want

Posted by Joseph j7uy5 on June 10, 2007
(4)
More »

I am
not skilled enough to do this, but maybe someone out there is.
 What I want is a utility that will take a bit of html, such
as for the sidebar in a blog, generate a image file that looks just
like the html would if it were rendered on a web page, then find all
the links and map them to that image file.

Then when a blog page loads, instead of calling up a gazillion little
picture files and running a bunch of little java scripts, it would just
need to load that one image.  It seems as though that would be
a lot faster.  Of course it should be smart enough to know
when it needs to regenerate itself, say, if you add something to your
blogroll.  But I would settle for one that has to be told when
to regenerate.  I suppose if it were part of the blogging
software, it could regenerate itself every day at 3AM or something like
that.

Some things would loose a bit, like those things that tally the cost of
the war in Iraq.  It would only change once a day.
 Some sidebar glitter, like the phase-of-the-moon or
current-weather things, would not be suitable.  You would have
to leave those as they are, or decide they are not worth the bandwidth
and get rid of them.  Not that I would ever tell anyone else
what they should do with their blog.  

(4)
More »

Comments

  1. #1 ukko
    June 10, 2007

    It sounds like you have got a bad DNS server or the ads are slowing down your connection. What you are describing would be bad in so many ways, The single image file would be much larger than the individual images and underlying html. It also would not be resolution independent, I am so happy that we have left the old days where every web designer thought that we were running at VGA resolution and could see a six pixel high font.

    Don’t even start thinking about how you would support the blind and other people with screen readers.

    Again, I think that you would be better off fixing the root cause of our troubles. The easiest of these to fix is DNS, just follow the directions at opendns.com and things wil improve.

  2. #2 Barry Leiba
    June 10, 2007

    It’s unlikely that DNS has anything to do with it. The problem is just that there’s a plethora of images, many from different places, and all the web servers that have to provide the images take a while to do it.

    While it’s true that one larger image takes longer to download than a few smaller images, Joseph has part of a good idea here: the bulk of the overhead in retrieving all this stuff is the session setup time, and setting up one session with one server to get one image, albeit a larger one, would be much faster than setting up, say, 50 sessions with 50 web servers and getting smaller images from them.

    But ukko’s right that it’s not really a good idea in the long run, because of layout issues. There’s also the complexity of mapping the links correctly, and the fact that you’d be turning all those nice text links into part of your graphic too.

    I think a better answer is this:
    Set something up that clones all the images to your own web server (in this case, the scienceblogs.com server). That’s easy to do, and it can be repeated at defined intervals to get the latest versions. Then change your sidebar so it gets the images from the scienceblogs.com server, and now all of our web browsers can use one keep-alive session to get all the images.

    Related alternative (more work for scienceblogs.com, but it does the above automatically):
    Have scienceblogs.com set up a way to proxy images. For instance, the “Daily Toon” image in this sidebar is from “http://www.andertoons.com/syndication/psychiatry_t.jpg”. Perhaps it could be set up to be retrieved as “http://scienceblogs.com/imageproxy/www.andertoons.com/syndication/psychiatry_t.jpg”. Again, the web browsers could take advantage of using one HTTP session to get multiple images, and the scienceblogs.com server could cache the images automatically, and only retrieve new versions periodically.

  3. #3 Joseph j7uy5
    June 10, 2007

    First of all, thanks for the tip on opendns. Previously I had suspected that problem and searched for “free dns servers” but did not find anything useful. It was a while ago (years) so maybe opendns was not up then. I did start using it and it seems to help. Hard to know without actually testing it, but at least it is no worse.

    Having said that, I did put the majority of the images right here on the scienceblogs server, and I still sit and watch them come up, a bunch at first, then a few stranglers one at a time. Some just time out. For example, right now I see that the “add to netvibes” graphic did not load. I can see in the status bar, messages “waiting for …” and I have to think there are significant delays at those times.

    The ads are an issue, but I use adblocker so I generally don’t see them. Sometimes I am nice to my host and whitelist this page, and click on all the ads.

    The utility would make it hard to implement accessibility features, unless the server could detect them and use the traditional means of drawing the page. Similarly, that could be done with “links” and other text-based browsers.

    Perhaps it would not help much. The only way to really know would be to try it and see.

  4. #4 Janne
    June 10, 2007

    If you’re just talking about aggregating that lump of aggregation icons then OK (though I would just get rid of them). But if you make text links – the whole sidebar – into an image you’re breaking things in several ways.

    First, anybody using an alternate browser technology that doesn’t support images, because they’re blind or for some other reason, will no longer be able to use anything in that sidebar. Second, you will break any browser that doesn’t support that size image for other reasons (like the small screen of a cellphone). Third, when the text links no longer behave like text links, that breaks a number of user interface assumptions for people – we tend to rely quite heavily on the changed color of visited links to keep tabs on where we’ve been already, for example.

    But of course, converting the text to an image saves you zero. That’s not where the bottleneck is at.