Forum OpenACS Q&A: patches vs upgrades

Collapse
1: patches vs upgrades
Posted by james shannon on 04/28/00 03:52 PM
Now that our project has reached public consumption, I think we have
to consider the few cases when it would be prudent to release a patch
instead of making people wait for upgrades. This would be in the case
of security problems that could compromise openacs or any other fatal
type error(probably none of which get to the public anyways).
Collapse
2: Response to patches vs upgrades (response to 1)
Posted by Ben Adida on 04/28/00 04:54 PM
This is a very good point. Yes, we should start thinking about patches, for sure. I do want to make one more full release, however, one that coincides with PostgreSQL 7.0 final and that includes a bunch of small tweaks and fixes, more documentation, the whole spiel.

In fact, if we could start doing some massive testing now for this FINAL OpenACS 3.2.2 release, that would be great. James, can you put together a plan of how you think we should go about releasing patches to be as community-friendly as possible?

Collapse
3: Response to patches vs upgrades (response to 1)
Posted by james shannon on 04/28/00 11:18 PM
Ok, this seems a little too simple, so I'm probably forgetting something... anybody? anybody?

  We should probably have a few of the most recent releases set up in various directories(not necessarily serving, just there). In that way, we can provide patches to people who are running a slightly non-current version of openacs that may still have the bug.

  After the bug has been discovered, fixed, and passes our litmus test for something that should be a patch(see below), we just run the diffs and set it up to work with the patch program(i've used this as a user, but never as a developer - shouldn't be too hard though).

  We then test and announce the patch on the website and the mailing list. Hopefully it will only be a single patch, but we may have several(depending on whether the problem files were updated across versions).

  Some notes: we should have a general idea of what will constitute a problem requiring a patch. As I'm writing this, I'm thinking of the sdm admin problem. Obviously little fixes will not require patches.

  I'm sure I've forgotten to mention something important. Comments? anyone? anyone?