You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by "David Bosschaert (JIRA)" <ji...@apache.org> on 2019/06/12 09:40:00 UTC

[jira] [Comment Edited] (SLING-8481) Caching of Feature Model Analyser Metadata

    [ https://issues.apache.org/jira/browse/SLING-8481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16861945#comment-16861945 ] 

David Bosschaert edited comment on SLING-8481 at 6/12/19 9:39 AM:
------------------------------------------------------------------

On using extensions to record this extra data. I think this could be a *lot* of data. IIUC then it could potentially be all the requirements and capabilities of all the bundles in the feature, for example. This would make the feature file very large. Not sure if this is a problem, but I just wanted to mention it.

Another question that I have about recording this caching data inside the feature file using extension is about identity. Feature models files are immutable. If you want to change them you need to create a different feature with a different identity. I'm not sure if that works for caching, because if you add this caching data to the feature after it was initially produced, the feature will have a different identity... And I'm not sure we want to prepopulate the cache eagerly either...


was (Author: bosschaert):
On using extensions to record this extra data. I think this could be a *lot* of data. IIUC then it could potentially be all the requirements and capabilities of all the bundles in the feature, for example. This would make the feature file very large. Not sure if this is a problem, but I just wanted to mention it.

Another question that I have about recording this caching data inside the feature file is about identity. Feature models files are immutable. If you want to change them you need to create a different feature with a different identity. I'm not sure if that works for caching, because if you add this caching data to the feature after it was initially produced, the feature will have a different identity... And I'm not sure we want to prepopulate the cache eagerly either...

> Caching of Feature Model Analyser Metadata
> ------------------------------------------
>
>                 Key: SLING-8481
>                 URL: https://issues.apache.org/jira/browse/SLING-8481
>             Project: Sling
>          Issue Type: Improvement
>          Components: Feature Model
>            Reporter: Dominik Süß
>            Priority: Major
>
> The feature model analyser is currently acting on the information of the feature model files as well as the metadata of artifacts. While feature models have a small footprint the extraction of the metadata from binaries is costly.
> Caching of this metadata should be improved in 2 ways:
> a) metadata should only be extracted once and be cached in process to avoid repetitive reading of binaries to access the same metadta
> b) it should be possible to store a "sidecar" artifact to a feature model to contain the extracted metadata from the binaries it references to eliminate the need for full binaries to be parsed or potentially even to have to be present for the analyser to run.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)