Support for eZ kernel 7.0 #122
Conversation
|
Any idea what else will be needed? Asking as it might not need a new major version unless it involves breaks. |
|
At the very least, support for the new HTTP cache bundle. I think that it's a breaking change, but not sure. |
| @@ -31,6 +31,7 @@ public function process(ContainerBuilder $container) | |||
| new Reference('ezplatform.http_cache.purge_client'), | |||
| new Reference('ezpublish.api.service.inner_content'), | |||
| new Reference('ezpublish.http_cache.event_dispatcher'), | |||
| new Reference('ezpublish.api.repository'), | |||
There was a problem hiding this comment.
this is also relevant for upcoming 6.7.7, 6.12.1 and 6.13.0.
Can code be done in different way to avoid having to set all arguments?
There was a problem hiding this comment.
There was a problem hiding this comment.
I'll try to find a way. This is just a playground for now :)
There was a problem hiding this comment.
I don't see a way to fix this any other way :/
The problem is that by this point any reference to old ezpublish.http_cache.purger.instant is removed from the container and we simply do not know the list of arguments.
| @@ -69,7 +69,7 @@ public function all() | |||
| return; | |||
| } | |||
|
|
|||
| $this->cache->clear(); | |||
| //$this->cache->clear(); | |||
There was a problem hiding this comment.
@andrerom I'm going to need some help with this one, or rather, with any commented out call to cache client here...
There was a problem hiding this comment.
Indeed, should clear by tag, will simplify this a lot and maybe also avoid some of the broad clearings.
well first we need to decide if we are going to have a branch only supporting 7.0 or have two adapters here being injected depending on version.
There was a problem hiding this comment.
I'm all for keeping it simple. 2.0 legacy bridge supporting only kernel 7.0.
If needed, this adapter could be backported than to 1.x.
There was a problem hiding this comment.
Ok, I'll push commit for this one.
There was a problem hiding this comment.
Don't forget other commented out calls, those are all I didn't know what to do with :)
|
@emodric pushed, not tested yet here, but on theoretical side only relevant todo left there is in regards to clearing type cache on type group changes. The user one is most likely not needed given how it is used. |
|
@andrerom Thanks a lot! It looks okay with only a reading of the code. We can always iterate later if something is missing :) |
emodric
force-pushed
the
2.0
branch
2 times, most recently
from
9e2801b to
ec3ded4
Compare
|
@andrerom In this state it is okay with me to merge. WDYT? I was thinking of removing |
|
So far so good, no issues. |
Just an unlock in
composer.jsonfor now, but obviously will need some more work :)