[gmx-developers] g87 support

[David van der Spoel]

On 2013-04-15 01:01, Mark Abraham wrote:
>
>
>
> On Sun, Apr 14, 2013 at 11:36 PM, Bogdan Costescu <bcostescu at gmail.com
> <mailto:bcostescu at gmail.com>> wrote:
>
>     On Sun, Apr 14, 2013 at 10:54 PM, Mark Abraham
>     <mark.j.abraham at gmail.com <mailto:mark.j.abraham at gmail.com>> wrote:
>
>
>         I propose we withdraw support for g87 for GROMACS 5, rather than
>         bother to fix it.
>
>
>     If it's broken anyway now, why not retiring it already in the next
>     versions of the current stables lines ? If someone would have missed
>     it, we'd have heard complaints already... Both 4.5 and 4.6 versions
>     are scheduled to be released very soon, that would make a perfect
>     occasion.
>
>
> If it's been broken for years, then there's no rush to withdraw the
> claim that we have working support for reading .g87, either. :-) There
> may be aspects of the support that do work that I failed to identify.
>
> In theory, our point releases are supposed to fix bugs and flesh out
> features that were not fully implemented in the major release. We
> haven't really talked about a feature-deprecation or feature-removal
> policy (though I have a number of candidates for deprecation for 5.0).
> Generally one would expect features to remain in a point release, but if
> they're already broken then the issue is fairly moot. If there's not
> time to fix something that would lead to non-obvious wrong results, then
> adding some kind of hack to save some users from disaster would be the
> way to go - but that's not applicable in the present case.
>
> We currently have no tests for trajectory-reading functionality, so any
> 4.x code change more invasive than "You want to read .g87? Sorry you
> can't have that. <fatal error>" runs a slight risk of breaking something
> else. One advantage of waiting for 5.0 is that we will have a proper
> test suite that would allow for making this kind of feature change with
> fair confidence.
>
> In theory, we're also giving someone the chance to say they really want
> .g87 - not that I imagine that person exists!
>
> Mark
>
>
There is one more aspect: .g87 is also the Amber ascii trajectory format 
(Peter Kollman borrowed a lot of code from Wilfred when was doing his 
sabbatical in Groningen). It would be good to hook up with Tom Cheathams 
trajectory reader (or merge that with our own efforts).

Anyway I haven't used this in a while but last time I did it still 
worked some way or another.


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se