Closed (fixed)
Project:
Search API
Version:
8.x-1.x-dev
Component:
Database backend
Priority:
Normal
Category:
Task
Assigned:
Unassigned
Reporter:
Created:
4 Aug 2015 at 15:09 UTC
Updated:
24 Aug 2015 at 09:04 UTC
Jump to comment: Most recent, Most recent file
Comments
Comment #1
LKS90 commentedHere is a patch, I used
$this->database->schema()->getFieldTypeMap()to get an array that kind of resembles the example in Schema.php. It fixed the issue in a really simple test that only had SearchAPI as a dependency but didn't actually do that much, therefore, feedback is welcome!Comment #2
LKS90 commentedComment #4
LKS90 commentedI think the ViewsTest fails are not related to this issue but this one: #2510564: Indexing is broken. No results are found and the page displays an error that complains about the
database_search_indexbeing unknown. The others are mostly accessing array indices that don't exist (anymore?).Comment #5
LKS90 commentedAlright, I update the $table each time I add a field and use the $table as a specification when adding indices.
Edit: This patch made the BackendTest pass for me, so it would be nice to get it tested :).
Comment #7
LKS90 commentedI fixed the schema errors. It's a tiny fix, so we can probably squeeze it in this patch. I fixed the schema errors by switching from
type: mappingtotype: config_entity, see the interdiff.Comment #8
LKS90 commentedComment #9
berdirThis should be config_object, actually, as it's not a config entity, just a plain config object.
Comment #10
LKS90 commentedSwitched to config_object.
Comment #11
drunken monkeyThanks a lot for reporting this issue and working on it, looks quite good already!
I've just made a few style changes, more or less, and then, since it passed for me locally, committed it. (I'm going on vacation tomorrow, for a week, so I thought it better to commit this right away.)
And yes, that test bot glitch really is annoying – not testing patches while HEAD is broken makes some kind of sense, I guess, but it's pretty stupid for actually fixing those fails.
It's
config_objectin this case, notconfig_entity. But interesting to see, didn't know about that change. Thanks for spotting this!Edit: Ah, sorry, had the post and patch prepared even before Sascha posted, but the tests took so long to run … :-/
Comment #13
berdirI guess that was a crosspost with the last two comments :)
Yeah, I can manually force patches to run on testbot with admin permissions, I usually try to do that in issues like this.
The problem with committing without that is that testbot will re-open the issue once testbot is green again and then this won't apply and will open the issue again :(
Comment #16
LKS90 commented