Nasty bug spotted: resolution in sweep block, -3d

It seems some of the refactoring I did recently left a bug lying around if you put a resolution in a sweep block e.g.


BEGIN PROJECT AUTOMATIC
BEGIN CRYSTAL DEFAULT

BEGIN WAVELENGTH NATIVE
WAVELENGTH 0.991870
END WAVELENGTH NATIVE

BEGIN SWEEP SWEEP1
WAVELENGTH NATIVE
RESOLUTION 1.7
DIRECTORY /Users/graeme/data/12847
IMAGE 12847_4_001.img
START_END 1 180
EPOCH 1106329906
BEAM 109.02 105.01
END SWEEP SWEEP1

BEGIN SWEEP SWEEP2
WAVELENGTH NATIVE
RESOLUTION 2.3
DIRECTORY /Users/graeme/data/12847
IMAGE 12847_5_001.img
START_END 1 90
EPOCH 1106332719
BEAM 109.04 105.01
END SWEEP SWEEP2

END CRYSTAL DEFAULT
END PROJECT AUTOMATIC

This has now bee squashed and will be mended in 0.3.4.1. Symptom is endless scaling loop talking about Resolution 1.0.

Manual in development

Busy writing a manual - if you think you would like to help debug it, please drop me a line. If you're really keen you can check it out of the xia2 subversion, it's in there.

xia2 0.3.4 release

xia2 0.3.4.0 is now available from the usual places, changes below. I would now suggest people get Aimless (and a new Pointless) from Phil Evans:

ftp://ftp.mrc-lmb.cam.ac.uk

and run -2da / -3da / -3daii

As always, any problems please drop an email to xia2.support@gmail.com

------------------------------

Changes since 0.3.3.4
 - Use Phil to override XDS parameters.
 - Mended issue on OS X Lion with default -parallel (was coming out at 1)
 - Use Pointless in place of Reindex; Aimless in place of Scala; no limit to number of batches which can be processed.

Changes since 0.3.3.3

 - Fixed bug in polarization correction vector for Pilatus data.
 - Fixed bug relating to assigning spacegroup on command-line.
 - Begun work on supporting Pilatus 300K for small molecule crystallography. Have now successfully processed data from Diamond Light Source beamline I19 EH2. N.B. Only works -3dii!

Just cooking up for a release - 0.3.4

Which should be a useful one... however sf.net appears to have disappeared for a moment so it'll have to wait. Hopefully today!

Where possible remove limitations

So, have been using Aimless for a little while now, which does not need recompiling for lots of batches. Using this with Reindex replaced with Pointless means you can now process data sets with many many images. By way of a test case: 70 180 degree sweeps @ 1 degree / frame from a small molecule crystal processed fine. This is most likely to be useful with Pilatus data. Would suggest -3da (and a local install of recent Pointless / Aimless versions) to do this kind of thing. Will be released shortly unless you want to get it from svn.

OS X Lion (ii)

OK, now the machine is working and crikey it works very well. First time I have had a laptop which is quicker at parallel processing than my 8-core Xeon workstation. Nice.

Unfortunately spotted some bugs in the xia2 implementation on there. Seems that they have changed the output on the program I used to guess the number of cpu's so they were not all being used. Now they will be.

Making OS X Lion work

Luckily just got my hands on a new and funky macbook pro with OS X Lion to help with user support. Here's how to turn it from a pretty paperweight into a computer (i.e. install compilers)

(1) install your updates to latest OS X
(2) get apple ID for app store
(3) install XCode from app store
(4) register as apple developer
(5) XCode -> preferences -> download (add command-line tools) -> install