On Nov 5, 2011, at 11:43 AM, Adrian Roitberg wrote:
> Computer language (and MD program you use) are best seen as a religion. You tend to program in the language you were taught in your 'household' and any hint of 'this is better than the other' is a simple bias. Bias is not a bad thing, but please do not say 'right or wrong language'. There is no such thing. They are tools to an end.
So we should start using a screw driver to pound nails? Your argument here is false. Let's just write amber in 100% assembly if the language doesn't matter. I'm not saying one language is superior to another in general, but for a given task it's completely clear that one tool is more appropriate than another and makes certain tasks easier. This isn't some language religion flame war.
>
> The interesting thing about dave's link is the comment there that ANY programer would ALWAYS say: I do not know who wrote the previous version, but I wish I can just destroy everything and start over in language XXX, which is clearly the right one for this work.
>
> Let's not fall into that.
>
Hence why I said that I don't necessarily suggest that we do it. It's clear that we have a tool that works, but there are some issues with extensibility and maintainability. At some point those issues might outweigh the pain of a re-write. I'm not sure that they do now, but the simple fact that every job I run now requires me to run three scripts in addition to tleap suggests that the right time might not be too far away. I'm not suggesting that it be rewritten in python just because I like python. I'm suggesting that _if_ it eventually needs to be scrapped and re-written from scratch it doesn't make sense to do it in C/Fortran/C++. Nothing more.
> At the end of the day, can we find two experience python programmers ?
>
> 'Maybe' yes.
>
> (what is 'experienced', would the next generation of 'amberites' agree with the 'experienced' choices or diss then?)
>
I have no idea what this means... I'm not sure how this became an experienced vs next generation fight with dissing...
> Can we find those and make sure they also are familiar with tleap and prmtop internal?
>
> I guess not
Probably not, but I mentioned four people.
>
> Can we find those AND make them stop what they are doing to pursue this?
>
> Surely not.
This might be the only point I agree with you on.
--
Justin MacCallum - Laufer Junior Fellow
Laufer Center for Physical and Quantitative Biology
Stony Brook University
justin.maccallum.me.com
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Nov 05 2011 - 09:30:03 PDT
