4.5.0: [20260822] Section editor: the categories of a section are now
those of the source layers, whole, instead of the ones the single
profile happens to cross. A class of the geological map that this
profile does not meet had no category in the section, so it could not
be given to a body drawn at depth, and the classification came out
different from one section to the next; the fallback colour, assigned
by the position a unit had in the list of that profile alone, also
gave the same unit a different colour in each of them. The whole
classification of the intersected layers - every class of the polygon
layer with the colour it is drawn in, every category of the
intersected lines - now travels with the export in a new
gp_source_categories table, the schema going to v4 with older
databases upgraded in place; an export written before it carries none,
and the categories are then gathered from all the loaded profiles
together, which still keeps them common to every section
[20260822] Section editor: a legend common to all the sections shown,
drawn in the plane beside them: a column of swatches, one per
category, in the order of the source layer legend and with the
category name beside it, in a "Legend" group holding one layer for the
geological units and one for the line intersections. It follows the
sections when the layout or a frame moves them, and is sized against
them. The new import option "Draw a legend beside the sections" leaves
it out
[20260822] Section editor: the styles of the categories are now shared
and kept. A change made to a category - on the legend, on the unit
bodies, on the outcrop bands or on the line intersections - is handed
at once to every other layer of its family, the bands taking the
colour of each category since they are lines where the bodies are
fills; and it is stored in the GeoPackage for the result set, so that
it is found again at the next load and in a later session, winning
over the attributes the export carries. The new "Reset the category
styles" button drops the choices made by hand and draws the categories
again as the export has them
[20260822] GeoProfiler and Section editor: the graphic attributes now
travel with the exported profiles, so a section is drawn in the
colours of the map it comes from. The export stores the styles the
plot has at that moment - the colour of every geological unit, the
colours, widths, line styles and markers of the topography and of the
projections, the elevation range and the vertical exaggeration - in a
new gp_graphical_params table, a profile exploded and restyled on its
own carrying its own alongside those of the stacked plot; the schema
goes to v3, older databases being upgraded in place and read as
before. The Section editor paints the outcrop bands and the unit
bodies seeded under them with the colour each unit has in the plot,
instead of a palette of its own assigned in alphabetical order, which
gave the same unit a different colour from one profile to the next;
the topography, the projected data and the line intersections keep the
styles they were plotted with. GeoProfiler puts the same styles back
when it reimports a geoprofile, which until now reopened with default
colours and recomputed axes - in a fresh session, with unit colours
drawn at random - and no longer resynchronises them with a polygon
layer the imported profiles have nothing to do with. The new import
option "Use the graphic attributes exported from GeoProfiler", on by
default and offered only when the database carries them, goes back to
the previous look
[20260822] GeoProfiler: fixed the export of a chosen subset of
profiles, which wrote no profile at all while reporting success: the
profiles were matched against the feature ids of the source line layer
through an attribute they do not carry, so every one of them was
skipped. The feature id each profile is drawn along is now kept with
the profiles - it cannot ride on the profiles themselves, draping the
traces on the DEM rebuilding them from their profilers - and reaches
the export, which until now also left source_profile_fid empty and
named every profile "profile" instead of after its source feature.
"Fill from selection" in the export dialog now also reads the
selection of a 3D line layer chosen among those already in the
project, and not only of one the plugin generated itself, the two
being remembered under different names
[20260822] Section editor: fixed the colour of the unit bodies, which
was neither the colour picked nor opaque as intended: the fill was
built by appending the opacity to the hexadecimal colour, an
eight-digit form Qt reads back as its own alpha-first one, so every
body was drawn in a shifted colour and at a sixth of the intended
opacity
[20260820] Section editor: the section is now framed: a box enclosing
the profile, an elevation grid, labelled ticks along the two axes and
the two profile ends named A and A', so that a section is read against
graduated axes instead of bare canvas coordinates. The frame is a
backdrop layer like the others, derived from the extent of the profile
alone and redrawn at every load: its two steps - elevation grid and
distance ticks - are chosen automatically, graduating each axis into
about eight round divisions, or set in the import dialog, and changing
one only takes loading the profile again. The box is grown outwards to
round elevations, and sits above the unit bodies, so the grid stays
visible through their fills. Nothing is asked of the GeoProfiler
export, which already carries the profile length and elevation range
[20260820] Section editor: the frame is not tied to the elevation
range of the topography any more: a bottom and a top can be given, in
the import dialog and in the new "Section frame" group of the panel,
together with the two steps, and "Redraw frame" draws the frame again
in place - no reload of the section, the interpretation untouched -
reporting the extent actually drawn, the automatic choices resolved.
The bottom is what a section usually needs: the box used to stop at
the lowest point of the topography, leaving everything interpreted
below sea level outside the grid. The given elevations only extend the
box, never cut the profile out of it, and the automatic grid step
follows the enlarged range. The four values are stored per profile in
the GeoPackage (gp_section_frame_params), so the frame comes back as
it was left, in a later session too; the frame options no longer
appear in the GeoProfiler import dialog, where they mean nothing
[20260820] GeoProfiler: fixed the symbolization of polygon
intersections, which did not follow the classification of the source
polygon layer: a class whose legend label is empty - the QGIS default
when the label is not filled in, as for the "all other values"
catch-all - was dropped, and its polygons were silently attributed to
the first class of the legend, which also received every feature the
classification does not cover; a layer classified by expression,
rather than by plain field, collapsed entirely into that first class,
its classification never being evaluated. Classes are now named after
their label, falling back on the value they classify, features outside
the classification form an "unclassified" class of their own instead
of joining an existing one, and the plot colours follow the same
names, rule-based and graduated renderers included, instead of being
read only from categorized ones. Polygon intersections of a
single-symbol layer, until now computed and then silently discarded on
export for want of a class name, are exported as "all features"
[20260820] GeoProfiler: the "unclassified" class, gathering the
polygons the source layer classification does not cover, now appears
among the polygon intersection colours, in a neutral grey that can be
changed like any other, instead of falling back on the plot default;
it is listed only when some intersection actually belongs to it, so
the legend never carries an empty class. The legend text size is
settable in the same dialog, renamed "Polygon intersection style", and
its default drops from the 10 points matplotlib uses to 7: the legend
sits in a narrow column beside the profiles, where the larger text
took room from the section itself
[20260820] Section editor: the unit bodies are no longer seeded one
per polygon of the source map: the bands of a same unit that touch one
another become a single body, the boundaries between the map polygons
they come from not being geological ones, and leaving them to be
merged by hand made the editing unmanageable on a map holding a unit
as many adjacent polygons; the same unit met again beyond another one
still gets a body of its own. Bodies seeded one per polygon before
this are recognized and not duplicated. The bodies also reach half as
deep below their outcrop band as they did, a starting shape being
easier to pull down than to push up
[20260820] new Section editor module: loads a profile from the
GeoProfiler export database into the QGIS canvas as a
distance-elevation (s-z) working plane, with read-only backdrop layers
(topography, projected attitudes drawn as segments dipping with their
apparent dip, projected points, line and polygon intersections) and
three editable GeoPackage layers - horizons, faults and unit bodies,
the latter seeded once per polygon intersection as its outcrop band
closed downward, to be reshaped into the interpreted body - carrying
geological attributes (type, movement sense, units, confidence) and
snapping to the attitudes and the topography; a plain-SQLite export
database is copied, on confirmation, into a GeoPackage twin that
stores the interpretation alongside the export tables, the backdrops
are stored there too and rewritten at every load, so a saved QGIS
project reopens complete instead of warning about temporary layers
that come back empty, and reloading the same profile restores the
interpretation drawn for it
4.4.0: [20260817] GeoProfiler: added projection of earthquake focal
mechanisms, plotted as beachballs either in map view or rotated into
the section plane, with their own style (quadrant colours, size,
outline, alpha, labels) and export/import to the profile database
[20260817] GeoProfiler: point and focal mechanism projection can now
read the vertical field as a depth instead of an elevation, measured
either from sea level (default) or below the topographic surface from
a DEM, in km (default) or m, converted to elevation for plotting
[20260817] GeoProfiler: projection of large datasets is orders of
magnitude faster (collinear segments of densified profiles are
collapsed, and points beyond the maximum distance are excluded by a
buffer pre-filter before any projection); the progress bar now
advances per block of points, over the whole run, instead of once per
profile
[20260817] GeoProfiler: fixed saving and loading of graphical
parameters, which failed on save and returned unusable parameters on
load; fixed projection along individual axes, which used the axis of
an arbitrary profile segment instead of the one given for each point;
fixed attitudes being paired with the wrong elevation when a record
had no valid one; made focal mechanism labels legible; removed
unreachable duplicate point-projection code
[20260817] GeoProfiler: a single profile can now be exploded into its
own window, with the full height to itself instead of a sliver of the
stacked plot, and with its own graphical parameters, restyled from the
window itself; several exploded profiles can stay open side by side
for comparison; fixed an exploded profile being labelled "1" and
taking the first profile's line-intersection colours whatever profile
it showed
[20260817] GeoProfiler: readable profile axes: vertical tick labels
without the always-zero decimal and capped in number (five per profile
by default), both axes optionally labelled in km, the vertical one
optionally as a depth counting downwards, and each axis named with its
quantity and unit; the profile number moved out of the way of the
elevation labels; all of it configurable in the graphical parameters
dialog and saved with the styles; fixed the requested elevation range
not being honoured, as writing out the ticks widened the axis to span
them
[20260818] fixed the reading of degenerate geometries, where a single
defective feature discarded the whole source layer: features with null
or empty geometries, and parts with too few points, are now skipped
and reported in the "qgSurf" log panel, in line and polygon
intersections, point/attitude/focal mechanism projections, plane
distances and best-fit plane; polygon intersections now keep the line
parts of results that are not lines only, as produced by a profile
touching a polygon; profiles defined from a line layer now read
single-part geometries as well, and no longer require multipart ones
[20260818] help: the GeoProfiler help is aligned to the current
interface (numbered command tree, focal mechanism projection with its
beachball styles, vertical values read as depths, single profile
exploded into its own window, axis units and names, polygon categories
taken from the layer symbology, unreadable features skipped and
logged) and illustrated again, with the figures to retake marked "TO
BE UPDATED" and the missing ones marked "TO INSERT"; same marking in
the stereonet help, whose interface, style and plot figures predate
the 4.3.0 additions its text already describes; note on skipped
records added to the best-fit plane and plane distances helps
[20260819] GeoProfiler: the profile number is now written inside the
top left corner of its plot box, clear of the elevation tick labels it
used to sit against, on a light patch that keeps it legible where the
topography reaches the corner; the elevation axis name moved closer to
the box, the room kept for the profile number no longer being needed
[20260819] GeoProfiler: generating a 3D profile without having defined
the output storage is now refused with an explanatory message, instead
of being reported as successful while nothing was written; a
generation whose lines cannot all be written to the storage layer now
reports what was written, instead of confirming success and confining
the reason to the "qgSurf" log panel
[20260819] GeoProfiler: the commands defining the profile storage now
say what went wrong: giving a new 3D line layer the name of one
already in the geopackage reported only "Error with adding vector
layer", and now names the conflict and points to "Append to existing
3D line layer in geopackage"; appending to a layer that does not
exist, or that does not hold 3D lines, is refused when the layer is
chosen instead of failing at write time, once the profile has been
computed; creating a new geopackage over an existing file, which
recreated it from scratch and destroyed every layer it held while
reporting success, now asks for confirmation first
[20260819] GeoProfiler: fixed focal mechanism and attitude projections
keeping one record per label: the projection stores one solution per
record key, and those two used the label alone as the key, so all the
records sharing a label value collapsed into the nearest one to the
profile - a layer whose label field is not unique showed a single
beachball, or a single attitude, per profile. Records are now keyed by
feature id as the point projections already were, and still labelled
as before; the exported "src_fid" of projected attitudes and focal
mechanisms, until now always empty, carries the source feature id
[20260819] GeoProfiler: the beachball outline width is now a style
setting, next to the outline colour, and its default drops from the 2
points obspy draws to 0.5: at the size beachballs have on a section, a
2 point outline blotted out the quadrant boundaries it was drawing
[20260819] help: the stereonet help is fully illustrated as well - the
interface, layer input, text input, plot style and plot figures are
retaken, and the two figures its text described without showing are
there: the P-T axes plot and the canvas-linked floating window with
its depth/Z histogram
[20260819] Stereoplot: fixed the crash on plotting after a change of
project ("wrapped C/C++ object of type QgsVectorLayer has been
deleted"): the input layer defined from the previous project was kept
and used although QGIS had destroyed it. The input is now checked
before every use - plot, redraw on selection change, depth histogram,
canvas-rectangle-linked replot - and a source whose layer is gone is
dropped and reported, asking for the input to be defined again
[20260819] help: the GeoProfiler help is fully illustrated - every
figure now shows the current interface or the current plot, the "TO BE
UPDATED" and "TO INSERT" markers are gone along with the legend box
that explained them, and the nineteen figures are numbered without
gaps; removed help_0.2.html, the illustrated guide the current help
replaced, which no code opened and whose captions described
screenshots since retaken under the same names
[20260819] help: the GeoProfiler quickstart no longer skips the
definition of the output geopackage and 3D line layer, which is
mandatory and comes first, and states that a topographic profile one
already has is used as it is, entering the workflow at "1 - Choose 3D
topographic profile"; quickstart, workflow diagram and section
numbering follow the numbered command tree
4.3.1: [20260803] fixed security issues flagged by the QGIS plugin
repository scanner (Bandit): replaced runtime assert statements with
explicit exceptions and silent except/pass blocks with logged
exceptions, in qgSurf's own code and in the bundled apsg/geogst/affine
libraries
4.3.0: [20260713] fixed a crash in the Plane Distances tool (DEM grid
CRS parameter)
[20260726] GeoProfiler: removed the separate "Create geoprofile" step
(each stage now computes immediately on confirmation); sped up
line/polygon intersection and point/attitude projection (up to ~240x
faster); fixed a point-projection crash and a topographic-profile
calculation bug; added configurable profile sampling distance; various
UI/compatibility fixes
[20260802] Stereoplot: added P-T axes and tangent lineation (Hoeppner)
diagrams with dedicated styles; added live canvas-rectangle-linked
filtering with a paired floating window showing the live stereoplot
and a vertical depth/Z histogram (local vs. total distribution); main
stereoplot now follows the layer's QGIS selection; fixed
invisible/missing direction arrows (tangent lineation, faults) and a
division-by-zero for near-vertical slickenlines; invalid records are
now skipped and reported instead of raising; GeoProfiler: default
export database format changed to SQLite
4.2.0: [20260601] porting to QGIS 4
4.1.1: [20260601] fixed security issues
4.1.0: [20260328] implemented export and import methods for
GeoProfiler; UI changes
4.0.2: [20260108] refactored embedded modules and fixed missing affine
link
4.0.1: [20260104] fixed not implemented feature and offset bug in
GeoProfiler
4.0.0: [20260101] new version of GeoProfiler with additional functions
3.2.2: [20241221] fixed error-print related bugs in Geoprofiler tool
3.2.1: [20240608] fixed mpl(>3.7.3)-related bug ("module
'matplotlib.cbook' has no attribute 'mplDeprecation'"); changed
experimental to False
3.2.0: [20231101] improved 'DEM-plane intersections' tool; renamed
'Points-Plane Distances' tool to 'Plane distances from points/DEM
cell' and improved
3.1.0: [20230615] fixed embedded mplstereonet and point-distances
bugs, in 'Best Fit Plane' added polygon layers as input and distances
from mean plane as output
3.0.1: [20230316] fixed np.float deprecation warning
3.0.0: [20221126] added profiles module; embedded mplstereonet;
restored original method for DEM-plane intersections; fixed various
bugs
2.1.3: [20220227] intercepted ogr Geometry Destroy action for
ogr-related bug
2.1.2: [20210310] updated embedded mplstereonet and apsg modules;
fixed macOs gdal import error
2.1.1: [20200502] fixed bugs related to (1) apsg plane plotting and
(2) plot of linear data
2.1.0: [20190312] in DEM-plane intersection, dip angle is decimal
(with one digit)
[20190501] in Geological Plane - Points Distances, more variables for
each point are calculated
2.0.0: added stereonet tool from geocouche; added module for
calculation of distance between points and a geological plane;
improved best-fit plane module;code refactoring; removed old modules
dependences
1.0.0: added pygsf; minor modifications to GUI and help for
best-fit-plane calculations
0.5.0: porting to QGIS 3
0.4.1: fixed bug related to deformation orientation; (re)added warning
on geographic DEMs, still not correctly supported
0.4.0: added support for geographic DEMs; fixed conceptual bug related
to the use of the project CRS
0.3.5: fixed bug related to the use of polar CRS (i.e., lat-lon)
0.3.4: removed (for QGis compatibility issues) both
"geosurface_simulation" and "geosurface_deformation" modules: they are
now an independent, stand-alone Python project (simSurf, available in
GitHub)
0.3.3: best-fit-plane: added point input from loaded point/line layer
0.3.2: best-fit-plane: added saving of one or more point set solutions
as shapefiles
0.3.1: support for on-the-fly-projection for Best-Fit-Plane
interpolation and DEM-plane intersection; minor bugs fixing
0.3.0: added modules for the processing of geological surfaces
(non-planar)
0.2.3: fix graphic bug (dem intersection symbol not deleted on plugin
close)
0.2.2: version compatible with QGis 2.0
0.2.1: calculate best-fit plane orientation given points drawn in the
QGis map window
0.2.0: map display/interaction is now through the QGis map window

None