Github user osma commented on the pull request:
https://github.com/apache/jena/pull/95#issuecomment-212280436
FWIW, I'm in favour of storing the result set before serialization, instead
of caching the serialized results as is apparently currently done. It makes
technical sense, for the reasons @afs mentioned. I don't care if it's slightly
slower than Varnish - if I want the extreme speed, I can always use Varnish as
an additional layer.
From a server administrator perspective, I'd prefer having some sort of
control over the cache size, preferably in terms of bytes rather than number of
items. Currently I have no idea how much this cache will affect Fuseki memory
usage. If it's, say, below 100MB currently, then I don't care that much, but if
it can get into the gigabyte range (for example with many large result sets),
then it's going to affect server performance and I'd perhaps rather allocate
that memory to Varnish (which can also store compressed responses, so the same
memory can fit many more objects).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
