Justin T Conroy
Justin T Conroy
A couple other observations which may or may not be relevant to this ticket. It might be appropriate to create additional issues for these: * If we move the "read"...
I added a call to `strm.sync()` immediately after the `strm.flush()` in the above example. This made no difference in the outcome.
I also tried the case of moving the read after the write (and the call to `sync()`), and that also is still failing.
Yes, I call `strm.seekg(0)` just before attempting to read with `strm.readsome(...)`. `strm.rdstate()` returns 0 after every operation on the stream. `strm.sync()` also returns 0. So, no errors.
@BrianPugh Any more insight on this issue?
Were you able to reproduce it with the code snippet I provided above?
@BrianPugh Is this still on your radar? It would be nice if we can get this resolved. Let me know if there's anything I can do to help.
Note that we were eventually able to work around this issue by using file pointers instead of streams. Using `fflush` and `fsync` works for persisting the files through reboots without...
Thanks, I'll report back here if we end up doing our own benchmarks, then.
I did, but I've been pretty slammed at work, so I haven't had time to share.