1lmfit-6.1, released 2dec15:
2
3- docs:
4  - new man page lmfit(7)
5  - 'BUGS' section in man pages
6- code cleanup:
7  - lmcurve_tyd in separate source file
8
9lmfit-6.0, released 24nov15:
10
11- API changes:
12  - a few more 'const' qualifiers in lmmin and lmcurve.
13  - new wrapper function lmcurve_tyd, for use in GROMACS
14- code cleanup:
15  - further unify code layout
16  - use temporary variable for readability in lmcurve (suggested by T. Klauser)
17
18lmfit-5.3, released 24nov15:
19
20- autotools:
21  - set default optimization -O3
22- monitoring:
23  - detect if norm is not finite, new outcome code 12 = "found nan"
24  - print parameters before invoking callback functiion
25- docs:
26  - correct typo in lmmin(3)
27- code cleanup:
28  - uniform code layout, starting from clang-format
29  - correct and improve comments, especially in qrfac
30  - uppercase matrix names
31- testing:
32  - new unit test directory test/, for development purposes only
33
34lmfit-5.2, released 23oct14:
35
36- bug fix:
37  - correct man page (5.0 API change)
38- acceleration(?):
39  - revert to just one malloc for all workspaces, now based on a char array
40
41lmfit-5.1, released 29aug13:
42
43- bug fix:
44  - FILE** initialization from &stdout was not standard compliant
45    (report by Jan Krieger, cf http://stackoverflow.com/questions/18505530);
46    now FILE* initialized to NULL
47- code cleanup:
48  - better wrapper for "extern C" declaration
49
50lmfit-5.0, released 9aug13:
51
52- API changes:
53  - removed callback function for printout
54  - removed control structure for printout; some par's go to lm_control_struct
55  - pivoting can no longer be switched off
56  - from now on, addition of par's to lm_control_struct is not an API change
57- bug fix:
58  - correct memory leak introduced by reversal of mallock trick
59  - restore default scale_diag=1
60- monitoring:
61  - improved text messages ("success" -> "converged", we should not say more)
62- demo: reworked examples
63  - new example nonlin1: solution of 2 nonlinear equations in 2 variables
64  - example surface1 improved and included in man page lmmin(3)
65  - otherwise testsuite started as new project fittest
66- packaging: configure now supports pkg-config
67
68lmfit-4.0, released 16jul13:
69
70- binary library renamed: lmmin -> lmfit
71- API changes:
72  - lmcurve.h no longer depends on lmmin.h, but only on new lmstruct.h
73  - low-level API lmdif is removed (code integrated in lmmin)
74  - lmcurve_fit -> lmcurve
75  - new control parameter 'pivot' (default: pivoting is used, as before)
76  - new structure lm_princon_struct, generalizing printflags
77  - in consequence thereof, new parameter lists for lmcurve and lmmin
78- code cleanup:
79  - ensure strict C90 compliance: no C++ comments; replace non-standard
80    struct assignement (reported by Elio Mazzocca, using LCC)
81  - revert malloc trick of release 3.4 (readability is more important)
82  - source code comment blocks moved around, to reduce duplications
83- build scripts cleanup:
84  - demo programs are no longer installed to $bindir
85  - configure.ac: cleaned up, and inserted comments
86- documentation:
87  - new web location: apps.jcns.fz-juelich.de/lmfit
88  - new entry page (dokuwiki) instead of man page
89  - man page split and largely rewritten: lmcurve(3), lmmin(3)
90
91lmfit-3.5, released 3feb13:
92
93- changed overall license to FreeBSD to facilitate inclusion in Linux
94  distributions (request by M. Hroncok)
95- directory doc renamed man
96
97lmfit-3.4, released 21jan13:
98
99- only one malloc call to allocate all arrays (proposal by F. Polchow)
100- liblmmin_la_LDFLAGS += -lm (required by Ubuntu, report by J. M. Agosta)
101- there is now an alternative ruby binding, using inline instead of swig
102  (contribution by I. Drozdov)
103- long message for degenerate case now suggests increasing epsilon
104  (suggestion by Nghia Ho)
105
106lmfit-3.3, released 12sep12:
107
108- moved "int" declarations to top of lm_printout_std() and lmmin()
109  (required by Visual Studio 2005: bug report M. Beek)
110- corrected typo 0.55 -> 0.5 (bug reports M. Kardas and D. Pescia)
111- man pages now produced by make; finally got rid of dmake
112- untabified code (hint by W. Pimenta)
113
114lmfit-3.2, released 28mar10 (preview 27mar10):
115
116- shared library is given correct version number (set in configure.ac)
117- man page lmfit(3) is copied by dmake to lmmin(3) and lmcurve_fit(3)
118
119lmfit-3.1, released 12mar10:
120
121- debian directory not part of tgz distribution
122- temporarily removed lmmin.3 (was symlink -> lmfit.3)
123
124lmfit-3.0, released 2mar10 (preview 27feb10):
125
126- for parameter close to zero, change step width for calculation of jabobian
127- outer iteration counter starting at zero (C style)
128- replace BUG by LMFIT_DEBUG_MESSAGES and .._MATRIX to allow control by -D
129- remove redundant calculation by exchanging workspaces and lm_lmpar output
130- eliminate unncessary wa1 = -wa1 computation
131- avoid passing wa3 to wa1, choose better names for work arrays in lm_par
132- migrate documentation from brain-damaged asciidoc to pod
133- rewrote interface documentation
134- API changes:
135  -- eliminate lm_initialize_control, provide defaults for float and double
136  -- separate limits and status records
137  -- separate generic minimization and one-dimensional curve fitting
138  -- simplify the latter
139- replace curve-fitting demo
140- provide surface-fitting demo (correspondence M. Rudolphi)
141- provide vector minimization demos
142- test for fnorm<=LM_DWARF, renumbered info values 0, 10, 11
143
144lmfit-2.6, released 29sep09:
145
146- nfev incrementation even if printout=0 (bug report C. Tanner)
147- correct t->tvec, y->yvec in man page  (bug report C. Tanner)
148
149lmfit-2.5, released 20sep09:
150
151- malloc==NULL ? (suggestion R. Strickland)
152- allow printout=NULL (suggestion J. Lodewyck)
153- improve infmsg (correspondence B. McHoull)
154- eliminate some static variables (p25, p5, p75, ..)
155- eliminate function data types; they obscure more than they help
156
157lmfit-2.4, released 27jul08:
158
159- clearer file headers
160- clearer licence (everything has ever been in the public domain)
161- machine-dependent arithmetic constants from float.h (suggestion M. Bydder)
162- minor improvements of comments
163
164lmfit-2.3, released 5apr08:
165
166- fixed bug sf#1817631: replaced confusing variable names in lmmin.c:lm_qrsolv()
167- accepted feature request 1814970: inclusion protection in headers
168- accepted feature request 1814976: cplusplus wrapper in headers
169- fixed bug sf#1868399: typos in debug messages
170- accepted proposal from M. Bydder:
171     improved calculation of differences in lmmin, lines 415-417
172- reverted comments in lmmin.c to pure C style
173- improved documentation of interfaces
174
175lmfit-2.2, released 20sep07:
176
177- no change log has been maintained,
178  there are just the sourceforge release notes
179