Currently, we throw watchdog errors in cases where no actual error exists. We need to re-implement what mollom() does with regard to returned error codes, or perhaps even think about whether we can re-use mollom() somehow.
| Comment | File | Size | Author |
|---|---|---|---|
| #10 | mollom-DRUPAL-6--1.getServerList.10.patch | 2.61 KB | sun |
| #8 | mollom-HEAD.getServerList.8.patch | 2.63 KB | sun |
| #5 | mollom-HEAD.getServerList.5.patch | 2.56 KB | sun |
| #3 | mollom-HEAD.getServerList.3.patch | 2.41 KB | sun |
| #2 | mollom-HEAD.server-list-retrieval-error-handling.patch | 532 bytes | dries |
Comments
Comment #1
sunThis is how I think it should work. mollom() already handles everything. Of course, the recursion is ugly, but I think it avoids to duplicate a lot of code and we get the identical fallback server handling for mollom.getServerList in the end.
Comment #2
dries commentedHere is an alternative -- not tested, just prototype code but I think it might work.
The advantage is that is keeps
mollom()clean.mollom()is already quite complex and the proposed patch in #1 makes it even harder to grok.Comment #3
sunIt is not really the error code that should make a difference. If we do not want to go with recursive mollom() attempt, then we'd have to do it like in attached patch. I hope the code comments explain it.
Comment #5
sund'oh. ;)
Comment #7
dries commentedIf we get a MOLLOM_REFRESH it is better to move to the next server instead of aborting, IMO.
If we get an error, we need to try the next server in the list. The error might be a network error (i.e. timeout because the server is down) so we should try the next server.
Comment #8
sunPerhaps this makes us happy? :)
Comment #9
dries commentedCertainly makes me happy. :)
Committed to CVS HEAD.
Comment #10
sunSame for D6, will test manually.
Comment #11
sunThanks for reporting, reviewing, and testing! Committed to D6.
A new development snapshot will be available within the next 12 hours. This improvement will be available in the next official release.