KNOWNBUGS


Known "bugs" in Elves 1.2.5 (current release)
found 12-11-04 unstable twotheta refinement in Wedger Elves
a sign error in the twotheta refinement step could make this process unstable.
The fix is to keep an eye on your twotheta value and perhaps keep resetting it to the known value until other refined parameters stabilize.

known 11-16-04 sortmtz crash affecting Scaler and Wedger Elves
the latest version of mosflm can include ultra-negative values in raw.mtz. This can confound sortmtz, which replaces missing values with what it beleives to be a large negative number.
The fix is to either use a more realistic resolution cutoff in Wedger, or add a VRSET card to the sorting scripts generated by Scaler Elves. This bug will be fixed in Elves 1.3, once a universally portable value of VRSET can be found.

Known "bugs" in Elves 1.2 (obsolete)

incompatibility with CCP4 5.0.x
this version of Elves did not work with the CCP4 5.0.x for a very very long list of reasons.

incompatibility with space group R3/H3 and R32/H32
this space group still causes confusion with some CCP4 programs (such as vecref in 5.0.2). However, I beleive I now have Elves 1.2.5 handling these space groups correctly with each program.

gawk bug on SGIs crashes background autoindexing.
a broken gawk that still ships with most SGIs was being used by Wedger Elves. This gawk creates a segmentaiton fault if indexing takes too long. Using nawk instead of gawk fixes this.

goodness of fit
the "goodness of fit" reported for autoindexing is misleading. Smaller values are better. Few have been confused by this, but this was changed to "distortion of fit" in the next version.

unstable twotheta refinement
the refinement of the value of twotheta was implemented in an unstable way. The fix is to keep an eye on your twotheta value and perhaps keep resetting it to the known value until other refined parameters stabilize.

NUSPOT OFF
the handling of the NUSPOT OFF workaround of a mosflm bug was not implemented properly. The fix is to add NUSPOT OFF manually if mosflm crashes with that QOPEN - no streams left error.

inf string crashes awk on OSF1
the awk program on OSF1 recognises the string "inf" as an overflow and instantly crashes. This makes any file containing these three letters "kryptonite" to awk on this platform. This includes calling one of your inflection point dataset "inf". The fix is to call your inflection point data set "infl", upgrading to Elves 1.2.5, or avoid using OSF1.

modulo 360 bug with phi jitter
breaks up wedges if headers have phi jitter a bug in the comparision of close phi values causes unrealistic breaks in wedges. The fix is to edit the erroneous wedges.txt file.

unhandled refscl/dsyev
this rare scala error sometimes appears under Linux. It usually appears when you only have one wavelength in Scaler, and the reference dataset is really the whole dataset. The fix is to avoid using the reference data set in scaling.

unhandled localscaling crash confounds Scaler
an error reported by localscaling is sometimes misinterpreted as coming from somewhere else. This can sometimes cause Scaler to give up for no good reason. The fix is to manually get localscaling running yourself.

bestFH.com crashes when anomalous data are missing
fix is to make sure you have DANO in your MTZ file, or upgrade to the latest bestFH.com

missing solvent content in Processer/Phaser
Processer can sometimes fail to find a reasonable solvent content. This results in Phaser defaulting to the default range of solvent contents, which is annoying if you went through the trouble to provide the molecular weight. Fix is to upgrade or run Phaser by hand with the right solvent content.

wrong space group in wARP after EPMR
the best-scoring space group from the EPMR runs is incorrectly applied to the data for all wARP runs. This is only a problem if the correct space group is not the best-scoring space group in the EPMR runs.

corrupted atom list in Phaser
the atom list in Phaser can sometimes be corrupted if the input mlphare script does not contain an anomalous occupancy. The fix is to just make sure you list an anomalous occupancy of zero if you don't have anomalous data.

Discovered 4-29-04 affecting Wedger Elves on OSX
Some macs had a problem reading the phi value from image headers. This is fixed in the current release after 5-5-04.

Discovered 5-4-04 affecting Wedger Elves on Debian
Wedger Elves would choose a broken version of awk as the default, and it would fail in critical image filename text processing calls. This is fixed in the current release after 5-5-04.


Known "bugs" in Elves 0.9.1 (obsolete)

Discovered 9-20-00

reindex.com (Phaser Elves) will crash if your initial mtz has a non-standard cell ordering. This is not a problem with the "default" Elves processing chain, but could crop up if you use reindex.com on mtzs generated by other programs. A fix is available here. It is still not known how to relate the reindexed cell to arbitrary fractional coordinates, so be careful!

Known 9-1-00

reindexing in merge.com (Wedger Elves) and sort_everything.com (Scaler Elves) is implemented differently that it is in reindex.com (Phaser Elves). This is not a problem with the "default" Elves processing chain, but could present problems if you use the merge.com or sort_everything.com scripts for other purposes. A fix will become available soon. In the meantime, just make sure all your Wedger runs use the same unit cell, and do your scaling in P222 if you don't know which of your axes are screws and rotations.


If you know of a bug that is not listed here, please see the Reporting Bugs Page.


Back to the Elves Page. 


This page is not finished. It will never be finished, and neither will yours. Admit it.
James Holton <JMHolton@lbl.gov>