Re: [Open Babel] (no subject)

[Geoff H. <ghu...@ws...>]

On Mar 20, 2004, at 9:10 AM, Michael Banck wrote:

>> We have converted it OK, but split it into smaller chunks first. 
>> Unless
>> there is zero memory leak I am sure this number of molecules in one 
>> file
>> could give a problem.

I have not yet found memory leaks. But Michael has a good point that 
this would be a nice stress test. I have a few other truly nasty tests 
I'm working on, so I'll add this one for the "bulletproofing" tests.

I expect though, that it's simply a RAM problem or an older version of 
Open Babel (and *not* the original babel 1.6!)

The current babel command-line program (and the previous one for that 
matter) assumed that it could load whole files into memory. In this 
particular case (a multi-molecule file), the whole set of molecules 
will be instantiated as an OBMolVector before the writing occurs.

I can attest though that trying it right now on Mac OS X and Open Babel 
1.100.2 works just fine:
(1.25GHz G4, 512MB RAM)

> [ghutchis@alumina] Downloads: ls -l nciopen3d.mol
> -rwxr-xr-x    1 ghutchis staff    179873013 Mar 20 20:13 nciopen3d.mol*

> [ghutchis@alumina] Downloads: time babel nciopen3d.mol nciopen3d.mol2
>
> real    0m17.634s
> user    0m9.130s
> sys     0m1.110s

Yes, that's right. We can zip through 172MB of NCI molecules in ~17 
seconds. The original babel is still grinding after several minutes.

So Gaokeng, what version of Open Babel are you using? Did you download 
the latest 1.100.2 release from http://openbabel.sf.net/ ?

Thanks,
-Geoff