You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
ESMF just uncovered a bug in the FieldGet() implementation that was introduced with 8.4.0. The bug uses an optional argument on the left-hand-side of an assignment, unprotected by if(present(..)). It is to be assumed that this bug can lead to memory corruption. Since FieldGet() is pretty ubiquitous in ESMF user code this needs to be addressed. We just started work on preparing patch release 8.4.1.
Interesting though to note is that none of the ESMF regression tests carried out for 8.4.0 (and since in the 8.5.0 development) have flagged, or failed due to the issue. We were just made aware of it by a user report of "strange failures".
ESMF just pushed a first beta tag for the 8.4.1. I think it would be worth switching to tag v8.4.1b01 since it will eliminate the memory corruption issue that we now know of in 8.4.0.
The text was updated successfully, but these errors were encountered:
@daniele-peano - @gold2718 and I are trying to put in place a more consistent workflow for CMCC. So issues should be raised for all bugs and enhancements that need to be done across the CMCC-CM system.
@mvertens the HSM people told me that version 8.4.1b01 is a development tag.
There are more recent tags: 8.5.0b11 and 8.5.0b12
but none of these is an official release.
Do we need to wait for a final release version before updating from ESMF v8.3.9 on Juno?
ESMF just uncovered a bug in the FieldGet() implementation that was introduced with 8.4.0. The bug uses an optional argument on the left-hand-side of an assignment, unprotected by if(present(..)). It is to be assumed that this bug can lead to memory corruption. Since FieldGet() is pretty ubiquitous in ESMF user code this needs to be addressed. We just started work on preparing patch release 8.4.1.
Interesting though to note is that none of the ESMF regression tests carried out for 8.4.0 (and since in the 8.5.0 development) have flagged, or failed due to the issue. We were just made aware of it by a user report of "strange failures".
ESMF just pushed a first beta tag for the 8.4.1. I think it would be worth switching to tag v8.4.1b01 since it will eliminate the memory corruption issue that we now know of in 8.4.0.
The text was updated successfully, but these errors were encountered: