rtipsy.pro: handle files larger than 2GB. #3
Conversation
|
So it still doesn't work. I tested and on these particular files (which are not extremely large) it seems that the header reading only works with "L", but still the end size is still very slightly off (1137874924 expected and 1137875008 is what it is). I have no idea what's going on with this... clearly a bigger issue is that L and LL seem like they are needed for different sized files which is extremely annoying. |
|
I can confirm that the header is being read correctly. The number of particles is accurate, but the sizes associated with each are off I think? Not sure why. |
|
On one of my runs (a larger one than the one where this is failing) it seems to work fine with "L" in the header and then comparing to "28LL+...". I am very confused. |
|
So in other words, the tipsy file from a smaller run is failing with the current fix and that of a larger file is working fine. |
|
For the file that fails, what does it report? That is, after, "RTIPSY ERROR: Header and file size inconsistent Estimates: Header bytes: 28 or 32 (either is OK)", what are the numbers that it reports, and what should those numbers be, given the numbers in the header? |
|
Here is an example: |
|
rtipsy works fine on the previous and next snapshots of the same simulation. This is rtipsy as it is right now in this branch. On all prior steps it works fine. I think it fails on the following steps 2 more times. This must have something to do with either 1) the number of particles just barely crossing some threshold or 2) some kind of rounding error? |
|
Would the problem file have been overwritten in a restart? |
File size calculations were done with 32bit integers, which won't work for large files. Changed to use 64 "LL" ints.