Ian Bicking: the old part of his blog

An Easier Legacy

Changing blog software can be hard, as many of us now doubt know. In a recent post Bill de hÓra discusses some of the issues exporting his existing blog to a new blog system. I know both the difficulty doing this sort of thing, the difficulties implementing it, and all the broken links left behind because of it. And it's all stupid. We can make things much easier if the people writing the software would just let go a little more and stop demanding that all the information go in some opaque (to the web) backend data silo (i.e., a database).

For blogs the transition story should be much easier. When you want to retire your blogging software you should rip your site. Spider the whole thing down. It's not that much space, even when you have lots of redundant files.

Of course, you probably can't serve the results up with Apache. Apache just... isn't good enough. For instance, your files may not have extensions; you have to keep some of the metadata, at least content type, and there's no clear way to do that in Apache. (There might be tricks or modules to do this in Apache, I'm not sure.) Also, the page names might be awkward, like /blog/?p=45 -- Apache interprets that sort of stuff, and we don't want to interpret it, we spidered it as a page (irregardless of what implements that page), and we just want to serve it up as a page. That's it.

But ignore the server portion for now -- we know we can build it if we want to. You now have an archive you can keep around forever and move to different hosts, and it doesn't rely on a software stack or any data formats except the one that matters most: the HTML that people read.

But it's not perfect. You have a link like /blog/2005/07/my-post, or even /blog/?p=45. You new system has a link like /blog/2007/01/another-post.html. They both share /blog/ -- you can configure your way around some (like ^/blog/200[012345]/), but for how long? Must you only change blog software on new years day, or on the first of the month?

When serving files you need a layered approach. First you look for this legacy content. If the content doesn't exist, you do the normal lookups, going to your new blogging software. The lookup must go to the full depth. That is, you don't look at /blog and decide what to dispatch to. You look at /blog/?p=45 (or whatever) and if you don't find it you start with the next option (the live software).

Then you only have to worry about overlaps. And that's not without concern; /blog/archive/ may in both systems. So some management may be necessary to rename or remove some of that legacy content.

This is a good solution for content that is timely, or archival. If you have content that is not timely, you are probably using a CMS, and you really want to move the content into the new system so you can continue to manage it in a consistent way.

In that case while it would be nice to keep URIs stable it might not be feasible. Which is why all stacks need easily managed redirects, with feedback taken from 404s in the logs.

A system setup like this still has some problems. How do you show all the old entries as well as the new on your archive page? While I'm not terribly concerned about the styling of old pages, sensible navigation would be nice -- if you have archive links in your old pages, it will appear as though you've stopped posting (when in fact that's just the last date when the page was updated). And you want to do searches across all your content.

These are also solvable problems, but they require some more significant changes. We need to move the data that is currently stored in the model out into the pages themselves. We need smart spidering, maybe building on Google Sitemaps or other enumerations of the "interesting" parts of a site. In addition notification of new content (by pushing out event notification, maybe in the form of APP). From there you could build an archive page entirely separate from the blog software, and capable of reading content from more than one location. Another more conservative option is just building some basic customizability into blog software so that you could copy the static HTML for your old archives, and the blog software would simply append this to the archive page.

For styling and navigation I think a pipeline approach is most appropriate -- we're working on Deliverance to do this this, but it's a tricky problem. For searching... well, that at least is already easy; reasonable hints along with a third-part search service should work well (where "reasonable hints" includes not having your archive page indexed), and if a third-party search won't work then privately-hosted search services are still likely to be of higher quality than searches built into blog software.

What annoys me is that technically none of this stuff is difficult (this is actually the dumbest/easiest way to handle legacy content), but existing stacks make it difficult. (I'm looking at you, Apache!) This kind of system design can be more stable than what we currently have, easier to maintain, more general, and just more web.

This is the kind of stuff I'm trying to tackle with Paste and WSGI and all these little bits of code fitting together. I might just be tilting at windmills in the endeavor, as developers both young and old just love their backend models and want to have total control over everything (driven in no small part by over-controlling customers, graphic designers, marketers, and their ilk). And I can't really expect WSGI to take over the world; it's not going to be the next Apache. But we are definitely going to give some of this stuff a serious try.

Created 07 Feb


This kind of simple HTTP based approach is definitely showing a lot of promise. It significantly simplifies scaling (look at Amazon's S3 service) but still leaves the difficult problem of maintaining transactional integrity -- which is inherently difficult to scale -- for the active part of your site.

We need to develop architectures that give us the easy scalability of simple HTTP for the majority of our requests while minimising that which is reliant on hard to scale transactional systems. The alternative is ever larger monolithic databases. Perhaps we can learn from Google. http://labs.google.com/papers/bigtable-osdi06.pdf


# Laurence Rowe

Somewhat interestingly, WebDAV has this idea of combining multiple requests (in an XML wrapper). The intention there is to do the entire set of requests in a transactional fashion. If you just have a way of referring to a transactional context, you could then explode that multirequest into a set of normal requests. This doesn't work across multiple backends (without, I suppose, some fancy transactional event system), but at least it offers transactions on some compound structures. (I've actually been thinking about this a lot as it relates to OHM, as it's kind of an outstanding issue there.)

The other approach is to be sloppy, and to be ready to deal with situations where the consistency of the system has been compromised. This is much more practical when you don't control all the pieces. Actually, it's about the only thing that is practical when you don't control all the pieces.

# Ian Bicking

Maybe you need something like _blog backup: http://www.doughellmann.com/projects/BlogBackup/

# karl