Richard Gowers

The MemoryReader is sort of a single chunk reader, but it won't work with very large files like that you're right

https://github.com/hpcanalytics/paper-hpc-py-parallel-mdanalysis this has a lot on I/O bottlenecks, maybe chunk reading would have helped here

So if the plane carrying all the mda devs crashed, it would be annoying if the package stopped working because we decided that numpy 1.(n+1) wouldn't work despite having not...

Yes. On Fri, Sep 16, 2022 at 00:18, Oliver Beckstein ***@***.***> wrote: > Sure. > > By now we seem to have settled on a typical pattern for dealing with...

@zemanj so with the openmp loops, I thought it was preferable to have as few fork/joins as possible, so ideally one for this double loop. Many it would be a...

I added tidying the namespace, we currently dump everything there, which isn't useful to users (doing tab completion)

I had a dig into this. I think this fix is probably fine, but I think there's a more fundamental issue with ChainReader not respecting the Reader base class, which...

Isn’t ts data the aux that is native to the Reader? On Thu, Aug 18, 2022 at 20:57, Hugo MacDermott-Opeskin < ***@***.***> wrote: > Is your feature request related to...

I've not got time to try this, but I think it might work if you added `if line.startswith('END'): models.append(pdbfile.tell())` into the code @orbeckst linked to above. This would add `END`...

Ew gross. So one option is to make some “reserved” topogyattrs statically typed. On Wed, Oct 23, 2019 at 16:07, Lily Wang wrote: > >>> gsd = mda.Universe(GSD) > >>>...