Make file-handling robust for errors

[GoogleCodeExporter]

Would request that a solver crash leaves the database in a state where the 
model can be re-opened using the -v flag. This would massively help with 
debugging.

Original issue reported on code.google.com by steve.br...@atkinsglobal.com on 8 Apr 2014 at 4:14

[GoogleCodeExporter]

I have just made a tiny modification. See whether this helps. Let me know.

Original comment by t.kozi...@gmail.com on 12 Apr 2014 at 5:46

• Changed state: Fixed

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

This one is going to be tricky, I think, but I will keep looking. I need to do 
this on Linux then, because it seems to work on Mac.

Original comment by t.kozi...@gmail.com on 14 Apr 2014 at 3:40

• Changed state: Accepted

[GoogleCodeExporter]

Now tested with clce748d1289 and I can confirm this hasn't fixed it - Viewer 
loads ok but on trying to show the first frame throws:

HDF5-DIAG: Error detected in HDF5 (1.8.10-patch1) thread 0:
  #000: H5A.c line 557 in H5Aopen(): unable to load attribute info from object
header for attribute: 'ints'
    major: Attribute
    minor: Unable to initialize object
  #001: H5Oattribute.c line 537 in H5O_attr_open_by_name(): can't locate
attribute: 'ints'
    major: Attribute
    minor: Object not found

Original comment by steve.br...@atkinsglobal.com on 15 Apr 2014 at 8:16

[GoogleCodeExporter]

If flushing the buffers does not solve, I may need to start closing and then 
re-opening files avery time I write. Rather dirty way of doing it. Seems, like 
Linux is less gentle about taking care of killed processes.

Original comment by t.kozi...@gmail.com on 15 Apr 2014 at 8:29

[GoogleCodeExporter]

Steve, see now. I have just committed a change, which I hope, will fix it. Let 
me know if it works. SOLFEC may be somewhat slower now, as files get reopened 
at every write, but it should not be hugely significant.

Original comment by t.kozi...@gmail.com on 8 Jun 2014 at 6:01

• Changed state: Started