Markus Pfeiffer
Markus Pfeiffer
Max Horn writes: > For our CI tests, it would be helpful if we could reliably and globally detect syntax errors -- these often are not detected by tests, yet...
Since I have been hacking on the input/output code, I just want to make 2 points: * For things like the Jupyter kernel (and other IDE features), it would be...
@fingolfin thanks for the hint, I wasn't aware of that.
Should add a test that at least tickles TSAN when run.
Should also measure performance implications. At the moment there is only a very small number of leasable builders/string available per thread, as opposed to the previously unbounded number *per transaction*.
For the record: simple performance tests did not show any effect of this change on performance.
Hi, I was trying to reproduce your issue and concocted the following javascript test: ``` db._drop("a"); db._drop("b"); db._drop("c"); db._createDocumentCollection("a"); db._createDocumentCollection("c"); db._createEdgeCollection("b"); averts = []; cverts = []; bedges = [];...
I can confirm that the above script exhibits the same erroneous behaviour on the `devel` branch.
As a workaround you can disable the `async-prefetch` optimizer rule, which prevents these errors from happening. For example in the javascript example above: ``` db._query(q, binds, { optimizer: { rules:...