Support/Comments for Provisioner 1.8-beta

This is an auto-generated topic for Provisioner 1.8-beta by shamblett.

Brief Description:

This package allows provisioning of a Revolution site from either a remote Revolution or Evolution site. Resources, elements, files, packages and user information can be viewed/imported. Please read the user guide in /assets/components/provisioner/docs for exact usage details. Also remember to install the revogateway connectors pack into any Evolution site you wish to use. See the changelog for details of fixes in this version. The source for this package is now in GitHub at http://github.com/shamblett/provisoner
Evolution site import tab added in this version.

Hi shamblett,

I have imported an Evolution 1.0.0 site in Chinese. Everything went well, except all the Chinese characters came out with strange symbols. I was wondering does Provisioner support utf-8 encoding?

My Environment
MySQL 5.1
PHP 5.33
Apache2
OS is Mac Snow Leopard

Yes, it should do, here’s a code snippet that shows the fields in a resource that get utf-8 encoded :-
as far as I know I do this to all text type fields in elements, files etc.

Where were you finding the corruption? I may have missed a field somewhere.

I’m trying to provision a Evo 1.0.2 site to a Revo 2.0.4-pl2 site, and I’ve run into two problems.

The first—and I’m fairly certain I can correct this myself—is that the "createdon" date for each resource is not being carried across when I do a site import. I was using "createdon" on the Evo site to date things like news releases and blog entries, but I should be able to set "published" to the same value as "createdon" in the MySQL database on the Evo site before I do an import, and that should resolve this issue.

The second issue I’m having is that resources with a "Resource Type" of "weblink" on the Evo are being imported into the Revo site with a "Class Key" of "modDocument". Is there some way to have them show up as "modWeblink" or "modSymLink" instead, depending on whether their content is simply a number (symlink) or something else (weblink)?

Looking at the MySQL databases again, it appears as though both "createdon" and "publishedon" are set for most resources in the Evo database, but in the Revo database, "createdon" has the same value for every resource, and "publishedon" has a value of 0 for every resource. As such, my idea of setting "publishedon" to the value of "createdon" so that Provisioner will bring that information across will not work. Maybe I can hack something together with a temporary TV...

The date thing is a bit annoying in that other people have said this in the past and various fixes have been tried to stop this, the thing about the evo import tab is that it basically does a ’select *’ query on the resource and sends it across, so these values whatever they are should be preserved and just recreated, I’ll raise another ticket on this and try and get to the bottom of it. I’m wondering if the XPDO save method is setting createdon to the saved data, thus making them all the same, hmm.

The symlink/weblink is a bug, I don’t check this and just create the resource as a modDocument, I’ll raise a ticket for this also, thanks.

I have tried few sites to test the latest 1.8 release:

1. Site is on 0963 (shared host), has about 250 resources. Everything works when imported selectively., but when using the new import tab, it fails to import some data, mainly the TVs and TV values. On different attempts, results are different. Not successful

2. The site is 1.0.2 (shared host), has about 160 resources, contains some non-english TV data (Urdu language), import was almost complete but Urdu content came in, in shape of ? ? ? ?.

3. The site is 1.0.4 (vps), has about 1200 resources, Nothing happens. But selective import work fine.

Also, should Provisioner keep the resource IDs same in the imported content?

All 3 sites are on different hosts.

related comments on github: http://github.com/shamblett/provisoner/issues#issue/23

Provisioner won’t keep the original id numbers as they are re-created on import so will get whatever MYSQL gives them.

Is it possible to mail the URL’s of these sites so I can import these into a test revo site to see what’s going on? My test sites work OK but obviously the real world is different. I need to trap the data coming back from these sites to see if its a gateway error, or Provisioner itself.

Quote from: shamblett at Oct 20, 2010, 04:49 AM

Provisioner won’t keep the original id numbers as they are re-created on import so will get whatever MYSQL gives them.
...

Ok, so can u suggest any easier way to track down the internal links and fix them?

For the sites urls, I can share 2 of them, will use ur @linux email.

I could look for the id’s as part of tag conversion and substitute them, I have the old->new map at the time so this is doable.