I have never understood the excitement behind Drupal install profiles. They purport to cut development time significantly and bring Drupal closer to the masses. Kris at The Worx Company calls them "a game changer for the entire CMS market." The patterns project has been long in development to provide platform to deploy install profiles to multiple sites. There's a ton of effort being expended every day on install profiles, and I think they're a waste of time. Here's why:
1. Virtually no two Drupal installs are the same
There is no use case where the effort spent developing an install profile is less than the effort spent installing and configuring the modules yourself because you're always going to have to heavily customize the profile (or the post-profile-install site). This is true for all but the very smallest installs.
2. Development is cheap; theming is expensive (for small sites)
A "blog" install profile that makes Drupal more Wordpress-y isn't going to drop in a custom theme for me. Development of brightbacon.com was 10% development/code and 90% theme. Until install profiles magically turn PSDs into Drupal themes, they aren't addressing the most expensive part of small site building.
3. Features are too custom to be usefully captured in install profiles (for complex sites)
For complex sites, there probably isn't an install profile out there that does what you need, or your needs are too specific to be captured in a useful install profile. Typical example: a Drupal site with Digg-like functionality. This became so complex that it's been forked off into its own distro (and for what it's worth, its namesake site, Drigg.org, has a crashed database table as of posting this entry).
I see a few ways that install profiles might be useful: Localized installs would be a good use. Setting up base Drupal installs with commonly-used modules would be nice; on any new Drupal install I instantly install Admin, CCK, Google Analytics, Image, ImageAPI, ImageCache, Mollom, Pathauto, Token, and Views modules -- a "base" install profile with these modules enabled would be helpful to have. But at that point why not just use a Drush script to do it, or even use Aquia Drupal? Or tarball my base install and use that every time?
Install profiles need to do a lot more before they're worth attention. Right now they're a neat demonstration of automation with very little business value. They're like umbrella hats -- a technical solution to a minor inconvenience that's more trouble than it's worth. Install profiles don't save enough time. They're not even that useful, let alone "game-changing."
Sun, 10/11/2009 - 04:24
What you argue may have been true in the past, but a forward looking assessment of install profiles is much different.
In Drupal 7, install profiles are modules and are thus much easier to write. Anyone who knows how to write a module can write an install profile. They also can have updates and interact with running sites, so you'll see more cases where the install profile heavily defines how your site functions, even after the install screen. Furthermore they can have updates, so you can have an install profile vendor who works to keep you up to date.
Furthermore, the real combination to look for is install profiles plue the features module. It will be possible in D7 with features to configure a site and use it to build and export a feature that is an install profile (this is mostly possible today), so it won't even take programming to build and share clones of sites.
Then you also have to consider the advent of Drush make, which can get and build software from anywhere on the net, even applying patches to existing code. With Drush make an install profile will be able to build extremely customized applications that have all the 3rd party requirement in place for you.
Finally, Boris Mann makes a great point that once you have distributions (like Open Atrium or Tattler) that have strongly defined functionality, better and more specific themes get created which will cut down on the theme customization time that you mentioned.
So in total, I think that the efforts putting into install profiles and the ancillary pieces such as features and Drush Make will allow profiles to be a game changing force in the CMS market. The one piece that I worry about is Drupal.org, and whether we will succeed in innovating enough with the packaging workflows and tools to support the depth of richness that Drupal itself has to offer.
Sun, 10/11/2009 - 04:25
The presence of one link to Boris Mann's site was enough to trigger the spam filter.
Mon, 10/12/2009 - 15:10
Sorry about that... it's Mollom! Too funny.
Post new comment