plonewars.com

I got (to date) two comments on how I structure plone instances, which prompts me to share some additional thoughts.

• Keeping a load of plone sites in one zope instance? That’s only doable if you’ve only got a stock plone. Perhaps one or two basic add-ons. As soon as you start to do custom development: separate those instances! It is just too much work to make sure your installer doesn’t mess up other sites (or that your configure.zcml doesn’t mess it up).
• There’s no need to keep all those instances in one homedir, though instancemanager makes it pretty easy to do that. If your company has its own server, why not create one homedir per project? At zest software we normally have one homedir per project. You can put everything in root-controlled /var/lib/* directories, but to me that’s not terribly attractive or needed. If you use your server only for hosting zope instances, in my I there’s nothing wrong with placing the instances in a homedir. Actually, it makes for handy instancemanager instance names: instancemanager update production, instancemanager preview --backup, etc.
• Not everything needs to be in a home directory. Instancemanager puts backups into /home/username/backups/instancename by default, but you can override that in a side-wide config to be something like /var/lib/zopebackups/username/instancename.
• To me, it is an absolute necessity to specify formally which add-on products and plone version make up an instance. I’ve had a few projects where I just zipped up the Products dir and used that as a bundle as I couldn’t reliably figure out which add-on product versions I used (plone 2.0.5 from a .tgz or copied from an svn version? ploneformmailer 0.3 beta or 0.3 final with some local modifications?).
• Something like instancemanager makes it real easy to test things out. Just copy your existing philips.py to philips3.py and swap plone 2.5 for plone 3.0 and start testing out that big customer project with plone 3.0. Easy as that. The easier you make that sort of stuff, the more you’re going to do it. It was a snap for me to test various cachefu versions with both plone 2.5 and 3.0 because I just had to copy one file and change one line.
• Setting up a well-working way of managing your production and development instances will take some time. I really hope the thoughts here (and in the previous article) help some people to get started, but you will have to find your own best way of working. And if you’re in a bigger company, you’ll have to make sure everyone can work with this. Most guys and galls at zest software use instancemanager as that helps getting everyone on track quickly, but the way we’ve structured things doesn’t inhibit Daniel (for instance) from setting up his own instance by hand ’cause that’s still easy to do with the bundles we’ve got set up. And a quick look at the instancemanager config gives you a readable list of add-on products and versions. No problems!

The mailing lists are the core of it all. There are a lot of different
mailing lists, but plone.users (for the generic usage questions) and
plone.dev (discussing the development of the core of plone itself) are
the most-used ones. Now, how to keep track of them? Here’s a few of
the things that I tried and thoughts that I have.

• Especially plone.users has a large volume. That necessitates a
  strategy. Not reading this list at all is an option, though not for
  me. Helping and getting help is important. Important for developers
  to learn the issues that baffle users; important for yourself if
  you’re stuck and need an answer or new search direction fairly soon;
  important for plone as a support channel. A quote from a mail to the
  plone.users mailing list that I saw coming in just a minute ago by
  Larry Pitcher:

  Thanks for your comments guys!
  One of the things I love about Plone is that so many of the core
  developers take the time to answer questions and write documentation.
  

• Having a load of mails from the mailing list end up in your inbox is
  a recipe for disaster. Either filter it out or, better, use one of
  the available mail-to-news gateways to read it via nntp. The choice
  is then whether you’ll read it directly on that news site (gmane,
  nabble, etc) or in a newsreader.

• I tried google groups, but I ditched it. I couldn’t send messages
  from my normal email address: google was determined to use my gmail
  account. Which meant I couldn’t post until I re-subscribed with my
  gmail address to all those mailing lists. Which I did, but it was too
  cumbersome in the end. Instead of a quick key combo to jump from new
  entry to new entry, I had to select the threads and the newsgroups
  by hand (=mouse). Ugh. That just took too much time.

• I very very very briefly tried nabble, but I couldn’t find any
  shortcuts there: even the individual messages had to be selected by
  hand. Perhaps I missed something. Well, it turns out that nabble is
  advertised on plone.org as an easy way for people to pose a question
  and look for answers without needing to subscribe to the full
  mailing list (and subsequently learn the meaning of the term
  "inundation"), not as a way for core developers to follow it all.

• I’m now back to the best-working combo I found out till date (which
  I used before trying google groups and nabble): gmane plus
  thunderbird. Gmane provides the nntp interface to the plone
  mailing lists and I use thunderbird to read it. The advantage, for
  me, over google groups is that I can just hit the space bar to go
  from new message to new message, quickly going through all the
  threads. I’m already noticing the difference as I’m about twice as
  active as in "my google groups period"

• My brother mentioned slrn as his newsreader of choice. I’ll leave
  it to him to write a praise-filled blog entry about it as I haven’t
  gotten around to testing that one out yet. I could use it as an
  excuse to try out old trusty pine again, though