[Setup] Will migration ever work?

[Nick Davis]

We've been spending a ridiculous amount of time trying to migrate from
2.0.5 to 2.1.1 .
This is a great shame because migration is a showstopper on what
otherwise looks like an exciting new release with more functionality and
much better performance.

I question whether the migration code was tested enough. It
should have been tested on a number of Data.fs s from many different
users. This would have revealed fragilities, and bugs would have been 
fixed before an official release, but I struggle to believe enough of 
this happened, because otherwise there wouldn't be such problems.

If anything from the core Plone has been customised in the existing
site, Plone 2.1.1 inevitably breaks without particularly useful error 
messages. Perhaps the migration procedure should first go looking for 
customisations and give warnings. With simply a list of changed files, 
the migration code could search for and flag up customised versions of
files that are in this list. Though we're not meant to customise through the
ZMI, most of the books unfortunately tell you initially to do this,
and I imagine most sites have had some customising through the ZMI at
some point because most people do what the books tell them, until later
realising that developing on the filesystem is better.
How many existing, not very technical, users, are going to be bitten by 
this?

Whats happened now is some of our Members folders are not converting 
properly, from PloneFolder to ATFolder etc. I've tried various things 
including deleting our products and bits of our site first. It seems 
other people have similar problems. I've tried Raphael Ritz's trick on 
my folders and most of them did convert, though this doesn't seem to 
work on the parent Members folder which is a Large Plone Folder. But 
that is no way to go about migrating a production system. There is 
always the chance of missing a folder and having a site in an undefined 
state.

To anyone who thinks their migration worked properly, I recommend
inspecting every directory on your site through the ZMI, to make sure it
is indeed an ATFolder not still a PloneFolder.

I know everyone is severely stretched on resources, and freely
gives of their time, and this release is a great achievement, but why 
release something which will cause people inevitably to run into 
problems like this? This is not a way to grow mindshare in Plone. Should 
this perhaps be described as experimental, not a stable release?

Would it have been better to stay with 2.0.5 and not release 2.1 at all, 
than release something that looks great but isn't really ready, and 
people on 2.0.5 spend ages struggling to migrate?

Please no-one take offence. I still like Plone despite the problems. I 
am sure Plone 2.1 is a great achievement and look forward to getting it 
working......er.....at some point in time.  ;-)

Nick