| Summary: | [PATCH] Make D* functions thread safe | ||
|---|---|---|---|
| Product: | POI | Reporter: | Patrick Böker <patrick.boeker> |
| Component: | SS Common | Assignee: | POI Developers List <dev> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | Keywords: | PatchAvailable |
| Priority: | P2 | ||
| Version: | 3.13-dev | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | All | ||
| Attachments: | dstar_threading.patch | ||
Just noticed that this bug also affects non-threaded but nested uses of a D* function. This patch fixes these cases also. This patch has proven reliable in our production system over the past two months. Is there any reason for not merging? |
Created attachment 32825 [details] dstar_threading.patch The DGet/DMin implementations relied on class variables. They thus unpredictably returned wrong values when using them in a multithreaded environment. This patch fixes this. The code is tested, but whether the multithreading issues I encountered are actually gone will only show in a week or two. It's a rather simple patch, so I'm quite confident anyways.