6:11 p.m.
Hi,
today I got an "out of memory" error from gretl, I think it's the first
time for me...
It happened when I did something like:
list L1 = L2 - L3
where L2 has 63 elements, and L3 has 13800 elements. That's because L3
has lots of duplicates -- this is of course a bug in my script and I
have corrected that. Nevertheless, given modern computers with Gigabytes
of memory, it still seems strange that gretl shouldn't be able to handle
that, no? Or am I missing something?
cheers,
sven
10:24 p.m.
On Wed, 7 Sep 2011, Sven Schreiber wrote:
today I got an "out of memory" error from gretl, I think
it's the first
time for me...
It happened when I did something like:
list L1 = L2 - L3
where L2 has 63 elements, and L3 has 13800 elements. That's because L3
has lots of duplicates -- this is of course a bug in my script and I
have corrected that. Nevertheless, given modern computers with Gigabytes
of memory, it still seems strange that gretl shouldn't be able to handle
that, no? Or am I missing something?
Well, in this case the error message is almost certainly misleading.
The "out of memory" flag is raised if the C-library memory
allocation function malloc() returns NULL. This can happen because
somebody has really asked for more memory than the system has
available, but it can also happen if, thanks to some bug, the amount
of memory requested was zero or negative. I suspect the latter in
this case.
If it's possible to create and post a minimal case of this failure
mode, that would be great.
Allin
10:43 p.m.
On Wed, 7 Sep 2011, Allin Cottrell wrote:
On Wed, 7 Sep 2011, Sven Schreiber wrote:
> today I got an "out of memory" error from gretl, I think it's the
first
> time for me...
>
> It happened when I did something like:
>
> list L1 = L2 - L3
>
> where L2 has 63 elements, and L3 has 13800 elements. That's because L3
> has lots of duplicates -- this is of course a bug in my script and I
> have corrected that. Nevertheless, given modern computers with Gigabytes
> of memory, it still seems strange that gretl shouldn't be able to handle
> that, no? Or am I missing something?
[...]
If it's possible to create and post a minimal case of this
failure mode, that
would be great.
Never mind, I have a test case and clearly it's a bug in calculation
of the memory needed. I should have a fix after band practice
tonight.
Allin
9:02 p.m.
On Wed, 7 Sep 2011, Allin Cottrell wrote:
Never mind, I have a test case and clearly it's a bug in
calculation of the
memory needed. I should have a fix after band practice tonight.
Now fixed in CVS. (Summary: trying to subtract list B with
duplicated elements from list A you could get a spurious "out
of memory" error.)
Allin
4867
days inactive
4868
days old
3 comments
2 participants
participants (2)
-
Allin Cottrell -
Sven Schreiber