Closed (fixed)
Project:
FeedAPI
Version:
5.x-1.0-beta3
Component:
Code (general)
Priority:
Normal
Category:
Support request
Assigned:
Reporter:
Created:
6 Jan 2008 at 05:10 UTC
Updated:
9 May 2009 at 04:56 UTC
This may be related to a few other issues already in the queue. See the attached screen shot of the errors.
What I did was add a vocabulary to the content type that is used for item node creation. Then I added terms to the feed node using freetagging. I saved it and everything appeared OK. Then I refreshed the feed and got the errors spewing all over the screen. Help!
I briefly looked through the code but the only thing that looked suspicious was the handling of updates. I seem to recall a comment stating something like 'this isn't really an update'.
This is on today's dev version and that of Jan 2nd.
| Comment | File | Size | Author |
|---|---|---|---|
| errorfeedapi.png | 77.16 KB | rconstantine |
Comments
Comment #1
rconstantine commentedActually, it looks like the nodes were updated with the new terms, so I don't know why those errors are there. With the Jan 2nd release, I was getting duplicate nodes created, but that seems to have been fixed with today's release.
Comment #2
jbowman commentedSorry, off topic, but in response to your mentioning that duplicate items aren't appearing anymore as of the latest version of CVS, does this mean now is a good time for me to test out multiple feeds getting the same item again, or was there a bug that caused multiple nodes from the same feed fixed?
Comment #3
aron novakI see that you use windows environment. I could not reproduce the problem under linux, now i'm installing xampp to my sandbox, stay tuned, i hope that i can catch the bug.
Comment #4
rconstantine commentedWhat I was getting with multiple nodes were from one feed only. Basically, I had created a feed using one of the many online feed-creators. When first created, the feed was OK, but the nodes that were made for each item didn't have the OG associations inherited from the parent feed. Refreshing the feed did the trick. But I later decided to add a vocabulary for feed item nodes to use, but which would also be inherited from the feed node. I added the vocab and it can be used just fine with other node types. Refreshing the feed once I add terms to the feed itself, however, was producing the error I originally mentioned, though the terms seem to have been added to the nodes.
I read elsewhere that old node would no longer be updated with OG and Taxonomy info, and I can understand why. Did it work in this case because the current feed's items matched those of existing nodes? I would very much like that if that is the case. That way, if we make changes to taxo' or OG stuff in the parent feed node, at least the current feed item nodes would reflect those changes.
Hmm. I'm curious about your concern about Windows. Do you have ideas that lead you to believe it could be the source of my problem?
Comment #5
aron novakin short: the problem is not specifically windows, but specific builds of php. the root of the problem is currently unknown, but at some php builds, we experience mysterious things: http://drupal.org/node/205706 . That's why i considered this possibility.
Comment #6
rconstantine commentedOne more thing to add is that I get the same errors when running cron from the site status page (admin/logs/status). Except of course, it is multiplied by the number of feeds I have. Cron does properly go back and associate OG data with the item nodes that should have been associated in the first place.
Comment #7
rconstantine commentedStill getting the errors with the latest beta. I'm wondering if you've updated the module to work with the taxonomy module from D5.5 and since I haven't updated yet that's why it isn't working... what do you think? Or have you not implemented code specifically to address the changes in that module?
Comment #8
rconstantine commentedWell I'm happy to report that Aron's assertion that things are fine on Linux is correct. However, my production server being Windoz, it would be nice to get rid of these errors. Plus, although web designers can often dictate the Web Host for clients, sometimes we can't and they might choose a Windoz package that we'd have to work with :-(
Comment #9
alex_b commentedComment #10
aron novakCan you provide me an exact software environment description of the production server where you experience this?
(anyway, i marked this critical, because it should be really fixed before 1.0 release)
Comment #11
rconstantine commentedWinXP, PHP 5.2.0, MySQL 5.0.27, Drupal 5.3 (I know, I know, I need to upgrade), Apache 2.0.59 (there was some reason I couldn't use 2.2; a mapping program IIRC)
Let me know what else you'd like to know.
Comment #12
aron novakIt's not about Windows. I tested it in a Windows (XAMPP) environment and I cannot reproduce the problem. Maybe it's an interaction with another 3rd party module. (as i spotted, you use pathauto too, for example)
Can you provide me a full list of enabled modules?
All in all, thank you for the bug report, while i tested this, i found another bug: feedapi enabled to set the same weight for two parsers/processors but the underlying logic could not handle this situation. Now feedapi validates the content-type form and prevents to setup nonsense values.
Comment #13
rconstantine commentedI'll update to the latest dev and make sure it's still happening. I don't suppose there isn't anything to do with weight of FeedAPI vs. Taxonomy like there was with FeedAPI and OG?
As for my list of modules, you are going to freak out. See the screen shot here: http://img338.imageshack.us/img338/8716/modulesnk6.png
Comment #14
aron novakSorry, i didn't write that i could fix your problem. it's still unknown for me. I suggest you to run feedapi automated tests in your sites: http://drupal.org/node/200037
Is it successful? If not, please paste the output here or there.
i suggest you to install a plain Drupal install with only og, feedapi and simpletest module in the same server environment, and see what happen. Trust me, 3rd party modules often makes their life impossible :(
Comment #15
Leeteq commentedComment #16
alex_b commentedClosing after an extended period of inactivity.