CVS log for ccp4/src/sfall.f
	Help

Request diff between arbitrary revisions / Display revisions graphically

merge to branch-30_07_11

added SAVE to LOGSAVE to prevent random crashes (Bugzilla #4545)

merge in 6.1.1 to 6.1.2 changes from the series-6_1 branch

increase format sizes for verbose output

Valgrind found uninitialised array

Valgrind found uninitialised array

Update to sfall to better handle formfactor names (Bugzilla #2890)

Fix from EJD. Problem from Stefano TRAPANI. Tidied by MDW.
SquAtmRadLimit is modified by BADD keyword.

Bug #1861: calculation of cos(alpha) squared appears to have been incorrect
in a number of subroutines, now fixed (see r1.175 in main branch).

Bug #1861: calculation of cos(alpha) squared appears to have been incorrect
in a number of subroutines, now fixed.

Increase MaxNumChains to 30 as requested by Phil

write symmetry info to output mtz file (bug 1447)

update licence banners

mixing of integer and character types - bug 1217

increase MCOLS to 500 in line with other progs

Make H,K,L,FP,SIGFP compulsory, as documented
Add protection against LOOKUP_SCR=0

change default to 5 gaussian scattering factor

Replace epsilon/centric/sysabs lot with CCP4SPG_F_* routines.
Ensures real spacegroup is used rather than fft spacegroup

Replace epsilon/centric/sysabs lot with CCP4SPG_F_* routines.
Ensures real spacegroup is used rather than fft spacegroup

update VDWR output

If there are MTZ file labels that duplicate the SFALL program labels then
stop with an error (otherwise the behaviour is unpredictable).

Fix to 1.160: need to declare SET_INDEX an integer

add new licence banners

changes from Clemens

allow longer spacegroup names

From EJD:
1) Use the cell associated with the dataset containing the FP column, and
   transfer the associated xtal and dataset names to the output columns.
2) Changed the default BFAC to 2.0 (in line with Refmac).

Allow BRESET 0.0

Check that MODE keyword given. Previously just checked if it was valid.

changes to allow MODE SFCALC MAPIN to run with defaults

slight correction to error trapping on XYZOPEN

Clarify error message if input data not sorted with l>=0

atomic -> formfac at request of EJD

changes to transfer crystal information from HKLIN to HKLOUT

Correct trap for MTZ file open failure.

oops - leftover from cut-and-paste in last change

New NAME keyword for assigning dataset for new MTZ file.

Make column assignment explicit (LRASSN) for scratch reflection
file. This is necessary for new library and neutral for old library.

correct error in format for loggraph (EJD)

small buglets

QREADX and QWRITX changed to QREAD and QWRITE

patch to make SFALL print the sum[a(i)] in the table of scattering  factors
Clemens V.

SFALL no uses the highest subclass of the primitive point group

should now recognise O   HOH water definition.

1) It wrote the wrong space group to the map header.
2) It didnt report anything about the map. EJD

MDW: Changes for possible wavelength in dataset header

Changes for new DCELL record in MTZ header.

Need to initialise resolution limits to prevent them being used before
they are set.

Increased MaxCoreSize. Added fix from EJD for error with unrestrained
refinement.

EJD: Reset gridsize for very high resolution.

Added calls to CENTPHASE in routines SOUTx to shift PHI to
nearest centric phase.

Standardised resolution label for loggraph use

Warning given if sort order in HKLIN is not h(slow)->l(fast). Changed
wording of error message in s/r SOUTZ.

1. Added traps in SGTEST, to give clearer messages about the need to
   increase MEMSIZE.
2. Added trap in SOUTZ, hopefully traps when indices on HKLIN are wrongly
   sorted. (Both prompted by Steve Ernst.)

Added stuff to cope with MTZ dataset info

initialising iter to 0 as xyzadvance was stopping at TER in pdb

Fixed recognition of atom names when reading in form factors.

Fix for refinement (from York).

a time saver from andrew mclachlan co EJD

EJD: changed the CHARACTER ElementNames*1
     Small fix to GENDEN for P1 structures; found by
     Andrew Machlachlan and Garib

Sort order now written to HKLOUT.
Plus some tidying....

Yet more changes to RWBROOK.

added new rwbrook stuff and put in 'normal termination' program terminators
through CCPERR.

EJD: someone with negative Fs!

EJD fix:   Someone has just managed to crash sfall when it is compiled
with the optimiser by having the first reflection of a block with a FP
flagged as MNF; somehow the LOGMSS got overwritten.

Moved SFSPACEGROUP stuff to one placce so it will be easier to change. Also,
improved automatic SFSP finding.

Bug fix from Eleanor.

reset_magic altered.

s/r SFREAD moved to rwbrook.f
EJD: GridSize SMN correction - defaults wrong..

Use CCPNUN to select file units. Close everything.
arrays(1) changed to scalars to help linux compilation (KDC)

ROTFIX changed to ROTPERM. Also, warning messages given only once.

Further work on missing number flag after problems with verbose.

EJD: 19-2-96   Someone was using sfall with a 3A cell edge! Disaster..
           But this check stops it failing..

Obsolete keywords removed. Scaling procedure changed so that missing refls
are removed. This means that results will not be the same as before. Also,
missing number stuff tightened up.

STEPSIZE variables didn't seem to be initialised properly.

Remove NGAUSS keyword which was recognised but a no-op.

EJD: obscure error found by someone abusing program

EJD: changed to read formfactors automatically if not specified! Still
no way to distinguisg Fe Fe+2, Fe+3 except to enter it yourself..

EJD Mod SFREAD to cope with comments in atomsf.lib.

Make `no reflections used' fatality in r3pi unconditional on i3

FREE column type should be I.
FREE keyword was no-op.
Fix dimensions to use MCOLS parameter everywhere.

EJD: move the calculation of NUNQ in R3PI outside conditional loop.
[Unknown effect.]

Avoid NaN lossage

EJD: checks connected with change in FREE keyword -- still don't have
doc.

EJD changes: Semantics of FREE keyword seem to have changed
-- need doc.  Missing data/free R stuff.

Remove un-useful o/p line.
EJD: initialise some variables.

10/4/95

RANGES could be undefined in previous change if not reading mtz file.
Also remove a few unused vars.

EJD: altered to flag mtz files whose reflections are in an incompatible
asymm unit for the SFSG requested.
I have also added subroutines to handle P3 sfs and refinement.

Initialise variable to avoid possible junk symmetry used if none input
when needed.

EJD change to `unexpected atomic number' message.
Also, previous change apparently only affects P4i2i2 SGs.

From EJD: correct errors in phase (not 0 or 180) for centric reflexions
h 0 l with h+l divisible by 3.

EJD:  Pointgroup name could be corrupted, which upset dm

Error message will be output if atom is unknown or ambiguous.

always unrecognised atom and residue names with verbose

After EJD:
change UNEXPECTED ATOMIC NO. diagnostic
`Atom name not recognised' diagnostic on with VERBOSE

EJD: If the number ofatoms > 20000,
the packing of Inumber + B does not work.. I think this will work

following EJD: extra SAVEing

avoid multiple SAVEs

don't allocate size for complex since datred needs array size for real

following EJD: fatal error with useful message when too many positions
+ `amino acid name not recognised' only with VERBOSE

typo

SAVEs

unused vars

Error trap for "ALLIN" when input columns are the same as program columns.
Note that when columns are entered explicitly no problem arises.

Remove unsued vars; use properly-typed diskio routines.
EJD: Nasty error in sfall when using FPART PHIPART; I also changed variable:
LFreeRflag to LSFrefFlag

Redoing P61/P65 changes.

Define ILOK a local parameter in KEYIN.

SAVEs; undefined variables in call to rbrook (from LMB).
Use EJD's method for defaulting SFSG rather than ours.  It should be changed
not to search for groups in its list.

see MTZLIB + 'ALLIN' now works?

Error from Eleanor in P61/P65, grid along z is now 12n.

Add title for hklout if no hklin.  Other trivia.

(Paul Cooper effort.)  Default value of SFSG to the highest symmetry one
that the program can do, consistent with the data.

Improve some error messages

correct bad SAVE

.

.

Add/change history info writing for a bunch of programs.  Mostly call the
new LWHSTL.  Several previously called incorrectly LWHIST with scalar
instead of character array.

Add history info to hklout

Save some local variables

insert SAVE, else problems with QSEEK in GENDEN

extra bit from york version to correct column label handling

EJD: problem with error estimate could cause failure

EJD: Error with ALLIN when NCOL > 29.
Also remove unnecessary SAVEs

Mainly remove multiple blank comments.

Mainly dynamic memory allocation.

Correct some scalars which should have been arrays and comment a dodgy
bit of integer/real casting.

Current version from York, much mangled by Kim

EJD: didnt cope with an mtz file with RESO 0

Reformatted by Kim, sigh, and with EJD changes.  The one I understand
avoids column-scrambling with free R and FPART input (only).

Mainly typed diskio calls.

FREERFLAG should have been changed to FREE

compilation error

change FRRERFLAG label to FREE, consistent with fft

EJD: MODEWT never initialised and problem if someone has no FREERFLAG.

.

PRE: changed line buffer size

.

.

small changes in step woth york

version pulled straight from york

missed density max fix in previous

change missed from EJD's

EJD: FreeR factor, NOSCAL if map input, Vctors stuff back.
presumably needs doc changing...

.

EJD: altered to stop people using both RMSSHIFT and STEPSIZE.

EJD: Problem if SMIN = 0.00

get calc of rho min and max right

EJD: Too mch verbosity

add ccprcs

EJD missing check using MODE SFCALC XYZIN   without an HKLIN file.

from EJD in response to:

From: OLIVA@BR.ANSP.USP.IFQSC
3) the option to disable the temperature factor dependent re-scaling of the
   data(RSCB) requires that DSMAX(=TSMAX) is set to 0.0, but this is checked
   within the program with something like "IF (TSMAX.NE.0.0) THEN...". Due to
   rounding errors, DSMAX (and TSMAX) are never exactly =0.0 and the check
   never works. I think that the option should be assigned to negative values
   of DSMAX, in which case a check of the type "IF (TSMAX.LT.0.0) THEN..."
   would be OK.

EJD: trans1 called with wrong no. of params

EJD: Change to sfall  for Garib: reset truncated values to TRNC0 and TRNC1

variable typo

EJD cosmetics

EJD Error in spacegroup P65 refinement

EJD: "for Judd"

rev. 1.33 change screwed

remove FACTRZ for fftlib

EJD: "change for Judd"

LMB: replace a check

EJD mods "for victor"

EJD: Victor pointed out STEPSIZE was not used
  when using MODE SFREF XYZB UNRESTRAINED

.

LMB typo in y'day's changes

LMB: Added check to number of Gaussians in sfall.for: wrong numbers
	gave JUNK STRUCTURE FACTORS! Also clarified man page: normally
	you shouldn't set NGAUSS (new sfall.for, sfall.1)

LMB: sfall change to table on
     +  '  $TABLE: Error analysis v B value ( very rough!) :',
where for arpn all atoms are waters - therefore no atoms
are summed for NB() as Carbons

       Average B =    38.52 ! removed this silly line

=================== if start file comes from pdbset then sfall fails
as all atosm are oxygen IZZ no carbons

put fndsmp in fftlib

remove setlim, setgrd from fft, sfall to fftlib

     Changes to P1 grid; make it divisible by 2 2 2;
  problems with changing map axes otherwise.

EJD: Damn: resolution checksdone if you are doing ATMMAP

previous patch wouldnt compile

EJD:    The check on grid factors in SFALL has ben reactivated.
   There are also some changes to make the program compatible with the FFT.

EJD: fix scaling for low resolution data

*** empty log message ***

EJD: correct correction

EJD: C2 structure factor generation is not excluding h+k = 2n+1.

EJD: Why hasnt sfall got theFree R factor stuff in?

small fix for consistency with york

EJD: It is wrong to cal sfall with SFREF XYZB RESTRAINED, and NOT
    have BATOM in prolsq input. Prolsq nowstops if this is so..

from EJD: fix bad array limits in soutn22

fix for rbfro1 from EJD

*** empty log message ***

initialise for rbfro1

tidying, variable defining from IJT

 From: bianchet@edu.jhu.med.cthulu (Mario Bianchet):
A small bug hides the label FC to the parser, affecting the use of the
Keyword SCALE FC in sfall.

EJD/PRE changes on branch 1.8.4.1.2
PRE comments:
I think if you let it choose a grid, it will probably get an
unsutiable one. But how to do it properly is not completely
straightforward. It is OK if you specify the GRID (for those options
where it is relevent).

updated from current LMB version

DL: fixed inline comments.  had inadvertantly altered some common blocks
relative to EJD's in last version.

EJD: Trying to fix fft SETGRD SETLIM stuff. Have changed FFT and SFALL to
  use same subroutines.
[this patch wrongly made to revision 1.8.2.2]

DL: fix up some common blocks

EJD: Trying to fix fft SETGRD SETLIM stuff. Have changed FFT and SFALL to
  use same subroutines.

    The patch below for sfall did lose precision for residue nos > 4000, so I
  have abandoned packing altogether and extended the BUF array from 11 to 12.
   Here are the diffs; prob best to pick up new one in a day or so
   affter the York guinea pigs have tried it.

EJD  This may lose precision better test...

EJD: Mods to allow more than 1000 residues per protein.
   Maybe this will work: just testing it.
   It should allow 10000 residues.

various standardisation

merged new EJD features with DL version previously fixed for portability

new version from EJD: new features:
    FREE R factor
    MODE SFREF XYZB UNRESTRAOINED possible now.
    RMSSHIFT XYZ ?? B ??  to limit unrestrained shifts..
    No need to give cell grid etc for
    MODE SFCALC MAPIN    all Taken from map.

typo in previous changes

from EJD: fixing Judd's scaling problem has ruined MY counter for map
correlation residue by residue!!
- genden: change nint to int
- writemap: no scale if counter set -- ir MODE=-5 or -6

`format fixes for Sheila' from EJD

fix for modewt from EJD

prior to proper attempt at version control

Roughly correposnds to first `official' release of MTZ version for un*x
and VMS around 14--15/5/92.

Initial revision