| 1 |
There is a problem with duplicate elements that needs to be handled. |
| 2 |
|
| 3 |
They can arise if someone uses conjoin to join all files on each |
| 4 |
specific processor and then uses epu to join the resulting files into |
| 5 |
a single file. If there is load-balancing, then an element can appear |
| 6 |
on multiple processors over time. |
| 7 |
|
| 8 |
Note that the element should only be active on 1 processor at any one |
| 9 |
time and its element variables are only valid on the processor on |
| 10 |
which it is active. Because of this, can't do a simple 'last one in' |
| 11 |
for the element values; need to pick which instance of the duplicate |
| 12 |
element is active at a specific timestep and then use that element's |
| 13 |
values for the output value at that step. |
| 14 |
|
| 15 |
Need something like "-remove_duplicate_elements" |
| 16 |
"-status_var <variable>" |
| 17 |
"-alive_value <value>" |
| 18 |
|
| 19 |
This means to eliminate duplicate elements and use the element where |
| 20 |
<variable> has value <value> as the "canonical" element at that time |
| 21 |
step. |
| 22 |
|
| 23 |
Probably throw an error if more than 1 of the duplicate elements has |
| 24 |
the value <value>... |
| 25 |
|
| 26 |
Note that this issue can be avoided if user epu's individual files |
| 27 |
first and then uses conjoin... |