the content tracker still needs to run synchronously. We need a mechanism to temporarily prevent text retrieval to avoid scalability issues and timeouts (caused by async transactions) - especially if we already know that they are long running like for ocr.
In the old, sync transformer framework it was possible to fake such a feature by setting cm:isContentIndexed=false to prevent the node to be catched up from the repository before it has been transformed and to remove that aspect later when the text transformation is available.
395 / SEARCH-2974 breakes this old "feature". So either we get a new feature to postpone the text retrieval or the mechanism for isContentIndexed is working again as expected e.g.
if a new node get's cm:isContentIndexed=false property added by behavior it must not result into an empty index doc
when removing the aspect or setting isContentIndexed to true later the text should be indexed
the content tracker still needs to run synchronously. We need a mechanism to temporarily prevent text retrieval to avoid scalability issues and timeouts (caused by async transactions) - especially if we already know that they are long running like for ocr.
In the old, sync transformer framework it was possible to fake such a feature by setting
cm:isContentIndexed=falseto prevent the node to be catched up from the repository before it has been transformed and to remove that aspect later when the text transformation is available.395 / SEARCH-2974 breakes this old "feature". So either we get a new feature to postpone the text retrieval or the mechanism for
isContentIndexedis working again as expected e.g.cm:isContentIndexed=falseproperty added by behavior it must not result into an empty index doc