| [ << ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
| 21.1 Timer Object | ||
| 21.2 XYPlot Object | ||
| 21.3 Canvas Object |
Timer objects can be used to make a timer that runs down toward 0 or runs up toward a pre-set value after which it starts blinking and returns itself to the application program. This can be used in many different ways, for example, to give a user a certain amount of time for completing a task, etc. Also hidden timer objects can be created. In this case the application program can take action at the moment the timer expires. For example, you can use this to show a message that remains visible until the user presses the "OK" button or until a certain amount of time has passed.
The precision of the timer is not very high. Don't count on anything better than, say, 50 milli-seconds. Run demo `timerprec.c' for an actual accuracy measurement.
| 21.1.1 Adding Timer Objects | ||
| 21.1.2 Timer Types | ||
| 21.1.3 Timer Interaction | ||
| 21.1.4 Other Timer Routines | ||
| 21.1.5 Timer Attributes | ||
| 21.1.6 Remarks |
To add a timer to a form you use the routine
FL_OBJECT *fl_add_timer(int type, FL_Coord x, FL_Coord y,
FL_Coord w, FL_Coord h, const char *label);
|
The meaning of the parameters is as usual.
There are at the moment three types of timers:
FL_NORMAL_TIMERVisible, Shows a label in a box which blinks when the timer expires.
FL_VALUE_TIMERVisible, showing the time left or the elapsed time. Blinks if the timer expires.
FL_HIDDEN_TIMERNot visible.
When a visible timer expires it starts blinking. The user can stop the blinking by pressing the mouse on it or by resetting the timer to 0.
The timer object is returned to the application program or its callback called when the timer expired per default. You can also switch off reporting the expiry of the timer by calling
int fl_set_object_return(FL_OBJECT *obj, unsigned int when) |
with when set to FL_RETURN_NONE. To re-enable
reporting call it with one of FL_RETURN_CHANGED,
FL_RETURN_END, FL_RETURN_END_CHANGED or
FL_RETURN_ALWAYS.
To set the timer to a particular value use
void fl_set_timer(FL_OBJECT *obj, double delay); |
delay gives the number of seconds the timer should run.
Use 0.0 to reset/de-blink the timer.
To obtain the time left in the timer use
double fl_get_timer(FL_OBJECT *obj); |
By default, a timer counts down toward zero and the value shown (for
FL_VALUE_TIMERs) is the time left until the timer expires. You
can change this default so the timer counts up and shows elapsed time
by calling
void fl_set_timer_countup(FL_OBJECT *obj, int yes_no); |
with a true value for the argument yes_no.
A timer can be temporarily suspended (stopwatch) using the following routine
void fl_suspend_timer(FL_OBJECT *obj); |
void fl_resume_timer(FL_OBJECT *obj); |
Unlike fl_set_timer() a suspended timer keeps its
internal state (total delay, time left etc.), so when it is resumed,
it starts from where it was suspended.
Finally there is a routine that allows the application program to
change the way the time is presented in FL_VALUE_TIMER:
typedef char *(FL_TIMER_FILTER)(FL_OBJECT *obj, double secs);
FL_TIMER_FILTER fl_set_timer_filter(FL_OBJECT *obj,
FL_TIMER_FILTER filter);
|
The function filter receives the timer ID and the time left for
count-down timers and the elapsed time for up-counting timers (in
units of seconds) and should return a string representation of the
time. The default filter returns the time in a
hour:minutes:seconds.fraction format.
Never use FL_NO_BOX as the boxtype for FL_VALUE_TIMERs.
The first color argument (col1) to
fl_set_object_color() controls the color of the timer,
the second (col2) is the blinking color.
Although having different APIs and the appearance of a different interaction behaviour, the way timers and timeout callbacks work is almost identical with one exception: you can deactivate a timer by deactivating the form it belongs to. While the form is deactivated, the timers callback will not be called, even if it expires. The interaction will only resume when the form is activated again.
See `timer.c' for the use of timers.
A xyplot object gives you an easy way to display a tabulated function generated on the fly or from an existing data file. An active xyplot is also available to model and/or change a function.
| 21.2.1 Adding XYPlot Objects | ||
| 21.2.2 XYPlot Types | ||
| 21.2.3 XYPlot Interaction | ||
| 21.2.4 Other XYPlot Routines | ||
| 21.2.5 XYPlot Attributes | ||
| 21.2.6 Remarks |
To add an xyplot object to a form use the routine
FL_OBJECT *fl_add_xyplot(int type, FL_Coord x, FL_Coord y,
FL_Coord w, FL_Coord h, const char *label);
|
It shows an empty box on the screen with the label per default below it.
The following types are available:
FL_NORMAL_XYPLOTFL_SQUARE_XYPLOTFL_CIRCLE_XYPLOTFL_FILL_XYPLOTData drawn as a solid line with the area under the curve filled.
FL_POINTS_XYPLOTFL_LINEPOINTS_XYPLOTData drawn as a solid line plus, per default, stars at data point.
FL_DASHED_XYPLOTFL_DOTTED_XYPLOTFL_DOTDASHED_XYPLOTFL_IMPULSE_XYPLOTFL_ACTIVE_XYPLOTData drawn as a solid line plus squares at data points, accepting manipulations.
FL_EMPTY_XYPLOTOnly the axes are drawn.
All xyplots per default display the curve auto-scaled to fit the plotting area. Although there is no limitation on the actual data, a non-monotonic increasing (or decreasing) x-axis might be plotted incorrectly.
XYPlots of type FL_POINTS_XYPLOT and
FL_LINEPOINTS_XYPLOT are special in that the application can
change the symbol drawn on the data point.
Only FL_ACTIVE_XYPLOT report mouse events by default. Clicking
and dragging the data points (marked with little squares) will change
the data and result in the object getting returned to the application
(or the object's callback getting invoked). By default, the reporting
happens only when the mouse is released. In some situations, reporting
changes as soon as they happen might be desirable. To control when
mouse events are returned use the function
int fl_set_object_return(FL_OBJECT *obj, unsigned int when); |
where when can have the folowing values:
FL_RETURN_NONENever return or invoke callback.
FL_RETURN_END_CHANGEDReturn or invoke callback at end (mouse release) if one of the points has been moved to a different place. This is the default.
FL_RETURN_CHANGEDReturn or invoke callback whenever a point has been moved.
FL_RETURN_ENDReturn or invoke callback at end (mouse release) regardless if a point has been moved is changed or not.
FL_RETURN_ALWAYSReturn or invoke callback when a point has been moved or the mouse button has been release).
Please note: an object can also be in inspect mode (see function
fl_set_xyplot_inspect() below). In this case the object
gets returned (or its callback invoked) for all of the above settings
except (FL_RETURN_NONE) when the mouse was released on
top of one of the points.
To obtain the current value of the point that has changed, use the routine
void fl_get_xyplot(FL_OBJECT *obj, float *x, float *y, int *i); |
where via i the data index (starting from 0) is returned while
via x and y the actual data point gets returned. If no
point has changed i will be set to -1.
It is possible tp switch drawing of the squares that mark an active plot on and off (default is on) using the following routine
void fl_set_xyplot_mark_active(FL_OBJECT *obj, int yes_no); |
with yes_no being set to false (0).
To set or replace the data for an xyplot, use
void fl_set_xyplot_data(FL_OBJECT *obj, float *x, float *y, int n,
const char *title, const char *xlabel,
const char *ylabel);
void fl_set_xyplot_data_double(FL_OBJECT *obj, double *x, double *y, int n,
const char *title, const char *xlabel,
const char *ylabel);
|
(The fl_set_xyplot_data_double() function allows to pass data
of type double but which get "demoted" to float type
when assigned to the xyplot object.) Here x, y is the
tabulated function, and n is the number of data points. If the
xyplot object being set already exists old data will be cleared. Note
that the tabulated function is copied internally so you can free or do
whatever you want with the x and y arrays after the
function has returned. title is a title that is drawn above the
XYPlot and xlabel and ylabel are the labels drawn at the
x- and y-axes.
You can also load a tabulated function from a file using the routine
int fl_set_xyplot_file(FL_OBJECT *obj, const char *filename,
const char *title, const char *xlabel,
const char *ylabel);
|
The data file should be an ASCII file consisting of data lines. Each
data line must have two columns, indicating the (x,y) pair with a
space, tab or comma separating the two columns. Lines that
start with any of !, ; or # are considered to be
comments and are ignored. The functions returns the number of data
points successfully read or 0 if the file couldn't be opened.
To get a copy of the current XYPLot data, use
int fl_get_xyplot_data_size(FL_OBJECT *obj); void fl_get_xyplot_data(FL_OBJECT *obj, float *x, *float y, int *n); |
The first function returns the number of data points which the second
will return. The caller must supply the space for the data returned
by fl_get_xyplot_data(). The last argument of that function is
again the number of points that got returned.
All XYPlot objects can be made aware of mouse clicks by using the following routine
void fl_set_xyplot_inspect(FL_OBJECT *obj, int yes_no); |
Once an XYPlot is in inspect mode, whenever the mouse is released and
the mouse position is on one of the data point, the object is returned
to the caller or its callback is invoked. You then can use
fl_get_xyplot() to find out which point the mouse was
clicked on.
Another, perhaps even more general, way to obtain the values from an XYPlot is to use a posthandler or an overlay positioner. See demo `xyplotall.c' for the use of posthandler and `positionerXOR.c' for an example of reading-out xyplot values using an overlayed positioner.
There are several routines to change the appearance of an XYPlot. First of all, you can change the number of tic marks using the following routine
void fl_set_xyplot_xtics(FL_OBJECT *obj, int major, int minor); void fl_set_xyplot_ytics(FL_OBJECT *obj, int major, int minor); |
where major and minor are the number of tic marks to be
placed on the axis and the number of divisions between major tic
marks. In particular, -1 suppresses the tic marks completely while 0
restores the default settings (which is 5 for the major and 2 for
the minor tic arguments).
Note that the actual scaling routine may choose a value other than
that requested if it decides that this would make the plot look nicer,
thus major and minor are only taken as a hint to the
scaling routine. However, in almost all cases the scaling routine will
not generate a major tic that differs from the requested value by more
than 3.
Normally the minor tics of logarithmic scales are drawn equidistant. To have them also drawn logarithmically use the functions
int fl_set_xyplot_log_minor_xtics(FL_OBJECT *obj, int yesno); int fl_set_xyplot_log_minor_ytics(FL_OBJECT *obj, int yesno); |
With this enabled e.g., the minor tics between 1 and 10 (when the interval is to be divided into 5 subintervals) will be drawn at the positions 2, 4, 6, and 8 instead of at 10^0.2, 10^0.4, 10^0.6 and 10^0.8. The functions return the previous setting.
It is possible to label the major tic marks with alphanumerical characters instead of numerical values. To this end, use the following routines
void fl_set_xyplot_alphaxtics(FL_OBJECT *obj, const char *major,
const char *minor);
void fl_set_xyplot_alphaytics(FL_OBJECT *obj, const char *major,
const char *minor);
|
where major is a string specifying the labels with the embedded
character | that specifies major divisions. For example, to
label a plot with Monday, Tuesday etc, major should be given as
"Monday|Tuesday|...".
Parameter minor is currently unused and the minor divisions are
set to 1, i.e, no divisions between major tic marks. Naturally the
number of major/minor divisions set by this routine and
fl_set_xyplot_xtics() and
fl_set_xyplot_ytics() can't be active at the same time
and the one that gets used is the one that was set last.
The above two functions can also be used to specify non-uniform and
arbitary major divisions. To achieve this you must embed the major tic
location information in the alphanumerical text. The location
information is introduced by the @ symbol and followed by a
float or integer number specifying the coordinates in world
coordinates. The entire location info should follow the label. For
example, "Begin@1|3/4@0.75|1.9@1.9" will produce three
major tic marks at 0.75, 1.0, and 1.9 with labels "3/4", "begin" and
"1.9".
To get a gridded XYPlot use the following routines
void fl_set_xyplot_xgrid(FL_OBJECT *obj, int xgrid); void fl_set_xyplot_ygrid(FL_OBJECT *obj, int ygrid); |
where xgrid and ygrid can be one of the following
FL_GRID_NONEFL_GRID_MAJORFL_GRID_MINORGrid for both the major and minor divisions.
The grid line by default is drawn using a dotted line, which you can change using the routine
int fl_set_xyplot_grid_linestyle(FL_OBJECT *obj, int style); |
where style is the line style (FL_SOLID, FL_DASH
etc. See section Drawing Objects, for a
complete list). The function returns the old grid linestyle.
By default, the plotting area is automatically adjusted for tic labels and titles so that a maximum plotting area results. This can in certain situations be undesirable. To control the plotting area manually, the following routines can be used
void fl_set_xyplot_fixed_xaxis(FL_OBJECT *obj, const char *lm,
const char *rm)
void fl_set_xyplot_fixed_yaxis(FL_OBJECT *obj, const char *bm,
const char *tm)
|
where lm and rm specify the right and left margin,
respectively, and bm and tm the bottom and top margins.
The pixel amounts are computed using the current label font and size.
Note that even for y-axis margins the length of the string, not the
height, is used as the margin, thus to leave space for one line of
text, a single character (say m) or two narrow characters (say
ii) should be used.
To restore automatic margin computation, set all margins to
NULL.
To change the size of the symbols drawn at data points, use the following routine
void fl_set_xyplot_symbolsize(FL_OBJECT *obj, int size); |
where size should be given in pixels. The default is 4.
For FL_POINTS_XYPLOT and FL_LINEPOINTS_XYPLOT (main
plot or overlay), the application program can change the symbol using
the following routine
typedef void (*FL_XYPLOT_SYMBOL)(FL_OBJECT *, int id,
FL_POINT *p, int n, int w, int h);
FL_XYPLOT_SYMBOL fl_set_xyplot_symbol(FL_OBJECT *obj, int id,
FL_XYPLOT_SYMBOL symbol);
|
where id is the overlay id (0 means the main plot, and you can
use -1 to indicate all), and symbol is a pointer to the
function that will be called to draw the symbols on the data point.
The parameters passed to this function are the object pointer, the
overlay id, the center of the symbol (p->x,
p->y), the number of data points (n) and the preferred
symbol size (w, h). If the type of the XYPlot
corresponding to id is not FL_POINTS_XYPLOT or
FL_LINESPOINTS_XYPLOT, the function will not be called.
To change for example a FL_LINEPOINTS_XYPLOT XYPlot to plot
filled small circles instead of the default crosses, the following
code could be used
void drawsymbol(FL_OBJECT *obj, int id,
FL_POINT *p, int n, int w, int h) {
int r = (w + h) / 4;
FL_POINT *ps = p + n;
for (; p < ps; p++)
fl_circf(p->x, p->y, r, FL_BLACK);
}
...
fl_set_xyplot_symbol(xyplot, 0, drawsymbol);
...
|
If a Xlib drawing routine is used it should use the current active
window (FL_ObjWin(obj)) and the current GC. Take care not to
call routines inside the drawsymbol() function that could
trigger a redraw of the XYPlot (such as
fl_set_object_color(), fl_set_xyplot_data()
etc.).
To use absolute bounds (as opposed to the bounds derived from the data), use the following routines
void fl_set_xyplot_xbounds(FL_OBJECT *obj, double min, double max); void fl_set_xyplot_ybounds(FL_OBJECT *obj, double min, double max); |
Data that fall outside of the range set this way will be clipped. To
restore autoscaling, call the function with max and min
set to exactly the same value. To reverse the axes (e.g., min
at right and max at left), set min > max for that axis.
To get the current bounds, use the following routines
void fl_get_xyplot_xbounds(FL_OBJECT *obj, float *min, float *max); void fl_get_xyplot_ybounds(FL_OBJECT *obj, float *min, float *max); |
To replace the value of a particular point use the routine
void fl_replace_xyplot_point(FL_OBJECT *obj, int index,
double x, double y);
|
Here index is the index of the value to be replaced. The first
value has an index of 0.
It is possible to overlay several plots together by calling
void fl_add_xyplot_overlay(FL_OBJECT *obj, int id, float *x, float *y,
int npoints, FL_COLOR col);
|
where id must be between 1 and
FL_MAX_XYPLOTOVERLAY (currently 32). Again, the data are copied
to an internal buffer (old data are freed if necessary).
As for the base data, a data file can be used to specify the (x,y) function
int fl_add_xyplot_overlay_file(FL_OBJECT *obj, int ID,
const char *file, FL_COLOR col);
|
The function returns the number of data points successfully read. The
type (FL_NORMAL_XYPLOT etc.) used in overlay plot is the same
as the object itself.
To change an overlay style, use the following call
void fl_set_xyplot_overlay_type(FL_OBJECT *obj, int id, int type); |
Note that although the API of adding an overlay is similar to adding an object, an XYPlot overlay is not a separate object. It is simply a property of an already existing XYPlot object.
To get the data of an overlay, use the following routine
void fl_get_xyplot_overlay_data(FL_OBJECT *obj, int id,
float x[], float y[], int *n);
|
where id specifies the overlay number between 1 and
FL_MAX_XYPLOTOVERLAY or the number set via
fl_set_xyplot_maxoverlays() (see below). (Actually, when
id is zero, this function returns the base data). The caller
must supply the storage space for the data. Upon function return,
n will be set to the number of data points retrieved.
Sometimes it may be more convenient and efficient to get the pointer to the data rather than a copy of the data. To this end, the following routine is available
void fl_get_xyplot_data_pointer(FL_OBJECT *obj, int id,
float **x, float **y, int *n);
|
Upon function return, x and y are set to point to the
data storage. You're free to modify the data and redraw the XYPlot
(via fl_redraw_object()). The pointers returned may not
be freed.
If needed, the maximum number of overlays an object can have (which by default is 32) can be changed using the following routine
int fl_set_xyplot_maxoverlays(FL_OBJECT *obj, int maxoverlays); |
The function returns the previous maximum number of overlays.
To obtain the number of data points, use the routine
int fl_get_xyplot_numdata(FL_OBJECT *obj, int id); |
where id is the overlay ID (with 0 being the base data set).
To insert a point into an xyplot, use the following routine
void fl_insert_xyplot_data(FL_OBJECT *obj, int id, int n,
double x, double y);
|
where id is the overlay ID; n is the index of the point
after which the data new point specified by x and y is
to be inserted. Set n to -1 to insert the point in front. To
append to the data, set n to be equal or larger than the return
value of fl_get_xyplot_numdata(obj, id).
To delete an overlay, use the following routine
void fl_delete_xyplot_overlay(FL_OBJECT *obj, int id); |
It is possible to place inset texts on an XYPlot using the following
routine (up to FL_MAX_XYPLOTOVERLAY or the value set via
fl_set_xyplot_maxoverlays() of such insets can be
accommodated):
void fl_add_xyplot_text(FL_OBJECT *obj, double x, double y,
const char *text, int align, FL_COLOR col);
|
where x and y are the (world) coordinates where text is
to be placed and align specifies the placement options relative to the
specified point (See fl_set_object_lalign() for valid
options). If you for example specify FL_ALIGN_LEFT, the text
will appear on the left of the point and flushed toward the point (see
Fig. 21.1). This is mostly consistent with the label alignment except
that now the bounding box (of the point) is of zero dimension. Normal
text interpretation applies, i.e., if text starts with @ a
symbol is drawn.
To remove an inset text, use the following routine
void fl_delete_xyplot_text(FL_OBJECT *obj, const char *text); |
Another kind of inset is the "keys" to the plots. A key is the combination of drawing a segment of the plot line style with a piece of text that describes what the corrsponding line represents. Obviously, keys are most useful when you have more than one plot (i.e., overlays). To add a key to a particular plot, use the following routine
void fl_set_xyplot_key(FL_OBJECT *obj, int id, const char *keys); |
where id again is the overlay ID. To remove a key, set the key
to NULL. All the keys will be drawn together inside a box. The
position of the keys can be set via
void fl_set_xyplot_key_position(FL_OBJECT *obj, float x, float y,
int align)
|
where x and y should be given in world coordinates.
align specifies the alignment of the entire key box
relative to the given position (see Fig.21.1).
The following routine combines the above two functions and may be more convenient to use
void fl_set_xyplot_keys(FL_OBJECT *obj, char *keys[],
float x, float y, int align);
|
where keys specifies the keys for each plot. The last element
of the array must be NULL to indicate the end. The array index
is the plot id, i.e., key[0] is the key for the base plot,
key[1] the key for the the first overlay etc.
To change the font the key text uses, the following routine is available
void fl_set_xyplot_key_font(FL_OBJECT *obj, int style, int size); |
Data may be interpolated using an nth order Lagrangian polynomial:
void fl_set_xyplot_interpolate(FL_OBJECT *obj, int id, int degree,
double grid);
|
where id is the overlay ID (use 0 for the base data set);
degree is the order of the polynomial to use (between 2 and 7)
and grid is the working grid onto which the data are to be
interpolated. To restore the default linear interpolation, use
degree set to 0 or 1.
To change the line thickness of an xyplot (base data or overlay), the follow routine is available:
void fl_set_xyplot_linewidth(FL_OBJECT *obj, int id, int width); |
Again, use a id of value 0 to indicate the base data. Setting
width to zero restores the server default and typically is the
fastest.
By default, a linear scale in both the x and y direction is used. To change the scaling, use the following call
void fl_set_xyplot_xscale(FL_OBJECT *obj, int scale, double base); void fl_set_xyplot_yscale(FL_OBJECT *obj, int scale, double base); |
where the valid scaling options for scale are
qFL_LINEAR and FL_LOG, and base is used only for
FL_LOG and in that case is the base of the logarithm to be
used.
Use the following routine to clear an xyplot
void fl_clear_xyplot(FL_OBJECT *obj); |
This routine frees all data associated with an XYPlot, including all
overlays and all inset texts. This routine does not reset all plotting
options, such as line thickness, major/minor divisions etc. nor does
it free all memories associated with the XYPlot, for this
fl_free_object() is needed.
The mapping between the screen coordinates and data can be obtained using the following routines
void fl_get_xyplot_xmapping(FL_OBJECT *obj, float *a, float *b); void fl_get_xyplot_xmapping(FL_OBJECT *obj, float *a, float *b); |
where a and b are the mapping constants and are used as
follows:
screenCoord = a * data + b (linear scale) screenCoord = a * log(data) / log(p) + b (log scale) |
where p is the base of the requested logarithm.
If you need to do conversions only occasionally (for example, converting the position of a mouse click to a data point or vice versa) the following routines might be more convenient
void fl_xyplot_s2w(FL_OBJECT *obj, double sx, double sy,
float *wx, float *wy);
void fl_xyplot_w2s(FL_OBJECT *obj, double wx, double wy,
float *sx, float *sy);
|
where sx and sy are the screen coordinates and wx
and wy are the world coordinates.
Finally, there's a function for returning the coordinates of the area of the object used for drawing the data (i.e., the area, when axes are displayed, which is enclosed by the axes):
void fl_get_xyplot_screen_area(FL_OBJECT *obj,
FL_COORD *llx, FL_COORD *lly,
FL_COORD *urx, FL_COORD *ury);
void fl_get_xyplot_world_area(FL_OBJECT *obj,
float *llx, float *lly,
float *urx, float *ury);
|
where via llx and lly the coordinates of the lower
left hand corner and via urx and ury those of the
upper right hand corner are returned. The first function returns
the corner positions in screen coordinates (relative to the object),
while the secoind returns them in "world" coordinates.
Don't use FL_NO_BOX as the boxtype of an XYPlot object that is
to be changed dynamically. To change the font size and style for the
tic labels, inset text etc., use fl_set_object_lsize()
and fl_set_object_lstyle().
The first color argument (col1) to
fl_set_object_color() controls the color of the box and
the second (col2) the actual XYPlot color.
The interpolation routine is public and can be used in the application program
int fl_interpolate(const float *inx, const float *iny, int num_in,
float *outx, float *outy, double grid, int ndeg);
|
If successful, the function returns the number of points in the
interpolated function ((inx[num_in - 1] - inx[0]) / grid +
1.01), otherwise it returns -1. Upon return, outx and
outy are set to the interpolated values. The caller must
allocate the storage for outx and outy.
See `xyplotall.c' and xyplotactive.c for examples of the
use of XYPlot objects. There is also an example program called
`xyplotover.c', which shows the use of overlays. In addition,
xyplotall.c shows a way of getting all mouse clicks without
necessarily using an active XYPlot.
It is possible to generate a PostScript output of an XYPlot. See the
function fl_object_ps_dump() documented in Part V.
A canvas is a managed plain X (sub)window. It it different from the
free object in that a canvas is guaranteed to be associated with a
window that is not shared with any other object, thus an application
program has more freedom in utilizing a canvas, such as using its own
colormap or rendering double-buffered OpenGL in it etc. A canvas is
also different from a raw application window because a canvas is
decorated differently and its geometry is managed, e.g., you can use
fl_set_object_resize() to control its position and size
after its parent form is resized.
You also should be aware that when using a canvas you'll
probably mostly program directly using basic Xlib functions,
XForms doesn't supply much more than a few helper functions.
You'll rather likely draw to it with Xlib functions and will
be dealing with XEvents yourself (instead having them
taken care of by XForms and cenverted to some simpler to
use events that then just return the object from
fl_do_forms() or invoke an associated callback
function. Thus you will typically need a basic knowledge of how to
program via the X11 Xlib.
| 21.3.1 Adding Canvas Objects | ||
| 21.3.2 Canvas Types | ||
| 21.3.3 Canvas Interaction | ||
| 21.3.4 Other Canvas Routines | ||
| 21.3.5 Canvas Attributes | ||
| 21.3.6 OpenGL Canvas |
Adding an object To add a canvas to a form you use the routine
FL_OBJECT *fl_add_canvas(int type, FL_Coord x, FL_Coord y,
FL_Coord w, FL_Coord h, const char *label);
|
The meaning of the parameters is as usual. The label is not drawn but
used as the window name for possible resource and playback purposes.
If label is empty, the window name will be generated on the fly as
flcanvasn, where n = 0, 1,....
The only types of canvases currently available is
FL_NORMAL_CANVAS.
The canvas class is designed to maximize the programmer's ability to
deal with situations where standard form classes may not be flexible
enough. With canvases, the programmer has complete control over
everything that can happen to a window. It thus doesn't work like
other objects that get returned by fl_do_forms() etc.
or have their callbacks invoked.
Instead the user can request that for specific X events (not
XForms object events like FL_PRESS, FL_KEYPRESS etc.!)
callbacks are invoked that receive all information about the
XEvent that led to their invocation. This obviously requires
some understanding of how the X Window system works.
The interaction with a canvas is typically set up as follows. First,
you register the X events you're interested in and their
handlers using the following routine
typedef int (*FL_HANDLE_CANVAS)(FL_OBJECT *obj, Window win,
int win_width, int win_height,
XEvent *xev, void *user_data);
void fl_add_canvas_handler(FL_OBJECT *obj, int event,
FL_HANDLE_CANVAS handler, void *user_data);
|
where event is the XEvent type, e.g., Expose etc.
The fl_add_canvas_handler() function first registers a
procedure with the event dispatching system of the Forms Library, then
it figures out the event masks corresponding to the event event
and invokes fl_addto_selected_xevent() to solicit the
event from the server. Other book keeping (e.g., drawing the box that
encloses the canvas, etc.) is done by the object handler.
When a canvas handler is installed the library tries to set the
correct mask for the the XEvent (which then tells the X Window
system which events to pass on to the Forms Library). But since
translation from an XEvent to an XEvent mask is not
unique, the default translation of the XEvent to a mask may or
may not match exactly the intention of the application. Two events,
namely MotionNotify and ButtonPress, are likely
candidates that need further clarification from the application. There
are two functions to add or delete from the mask,
fl_addto_selected_xevent() and
fl_remove_selected_xevent().
By default, when a mouse motion handler (i.e., for the
MotionNotify events) is registered, it is assumed that, while
the application wants to be informed about mouse movements, it's not
interested in a continous motion monitoring (tracking), thus per
default MotionNotify events are requested with
PointerMotionHintMask being set in the mask to reduce the
number of events generated. If this is not the case and in fact the
application wants to use the mouse motion as some type of graphics
control, the default behavior would appear "jerky" as not every mouse
motion is reported. To change the default behavior so that every mouse
motion is reported, you need to call
fl_remove_selected_xevent() with mask set to
PointerMotionHintMask. Furthermore, the mouse motion is
reported regardless if a mouse button is pressed or not. If the
application is interested in mouse motion only when a mouse button is
pressed fl_remove_selected_xevent() should be called with
a mask of PointerMotionMask|PointerMotionHintMask.
With ButtonPress events you need to call
fl_addto_selected_xevent() with a mask of
OwnerGrabButtonMask if you are to add or remove other canvas
handlers in the button press handler.
To remove a registered handler, use
void fl_remove_canvas_handler(FL_OBJECT *obj, int event,
FL_CANVAS_HANDLER handler);
|
After this function call the canvas ceases to receive the events for
event. The corresponding default bits in the XEvent mask
as were set by fl_add_canvas_handler() are cleared.
If you added extra ones with fl_addto_selected_xevent()
you should reset them using fl_remove_selected_xevent().
To obtain the window ID of a canvas, use
Window fl_get_canvas_id(FL_OBJECT *obj); |
or use the generic function (macro) (recommended)
Window FL_ObjWin(FL_OBJECT *obj); |
Of course, the window ID only has a meaning after the form/canvas is
shown. When the canvas or the form the canvas is on is hidden (via
fl_hide_object() or fl_hide_form()), the
canvas window may be destroyed. If the canvas is shown again, a new
window ID for the canvas may be created. Thus recording the canvas
window ID in a static variable is not the right thing to do. It is
much safer (and it doesn't add any run-time overhead) to obtain the
canvas window ID via FL_ObjWin() whenever it's needed. If
your application must show and hide the canvas/form repeatedly, you
might consider to "unmap" the window, a way of removing the window
from the screen without actually destroying it and later re-mapping
the window to show it. The Xlib API functions for doing this are
XUnmapWindow() and XMapWindow(). Both require two
arguments. the display, which you can determine by calling
fl_get_display() and the window ID, which can be obtained
by using form->window if you want to (un)map a form or
FL_ObjWin(obj) for a canvas.
Upon canvas creation, all its window related attributes, e.g., visual, depth and colormap etc., are inherited from its parent (i.e., the window of the form the canvas belongs to). To modify any attributes of the canvas, use the following routine
void fl_set_canvas_attributes(FL_OBJECT *obj, unsigned mask,
XSetWindowAttributes *xswa);
|
See XSetWindowAttributes() for the definition of the structure
members. Note that this routine should not be used to manipulate
events.
Other functions exists that can be used to modify the color/visual property of a canvas:
void fl_set_canvas_colormap(FL_OBJECT *obj, Colormap map); Colormap fl_get_canvas_colormap(FL_OBJECT *obj); void fl_set_canvas_visual(FL_OBJECT *obj, Visual *vi); void fl_set_canvas_depth(FL_OBJECT *obj, int depth); int fl_get_canvas_depth(FL_OBJECT *obj); |
Note that changing visual or depth does not generally make sense once the canvas window is created (which happens when the parent form is shown). Also, typically if you change the canvas visual, you probably should also change the canvas depth to match the visual.
Caution should also applied when using
fl_set_canvas_colormap(): when the canvas window goes
away, e.g., as a result of a call of fl_hide_form(), the
colormap associated with the canvas is freed (destroyed). This likely
will cause problems if a single colormap is used for multiple canvases
as each canvas will attempt to free the same colormap, resulting in
an X error. If your application works this way, i.e., the same
colormap is used on multiple canvases (via
fl_set_canvas_colormap()), you should use the following
routine to prevent the canvas from freeing the colormap:
void fl_share_canvas_colormap(FL_OBJECT *obj, Colormap colormap); |
This function works the same way as
fl_set_canvas_colormap() except that it also sets a
internal flag so the colormap isn't freed when the canvas goes away.
By default, canvases are decorated with an FL_DOWN_FRAME. To
change the decoration, change the the boxtype of the canvas and the
boxtype will be translated into a frame that best approximates the
appearance of the request boxtype (e.g., a FL_DOWN_BOX is
translated into a FL_DOWN_FRAME etc). Note that not all frame
types are appropriate for decorations.
The following routine is provided to facilitate the creation of a colormap appropriate for a given visual to be used with a canvas:
Colormap fl_create_colormap(XVisualInfo *xvinfo, int n_colors); |
where n_colors indicates how many colors in the newly created
colormap should be filled with XForms' default colors (to avoid
flashing effects). Note however, that the colormap entry 0 is
allocated with either black or white even if you specify 0 for
n_colors. To prevent this from happening (so you get a
completely empty colormap), set n_colors to -1. See section Drawing Objects, on how to obtain the
XVisualInfo for the window. Depending on the window manager, a
colormap other than the default may not get installed correctly. If
you're working with such a window manager, you may have to install the
colormap yourself when the mouse pointer enters the canvas using
XInstallColormap().
By default, objects with shortcuts appearing on the same form as the
canvas will "steal" keyboard inputs if they match the shortcuts. To
disable this feature, use the following routine with a false (0)
value for yes_no:
void fl_canvas_yield_to_shortcut(FL_OBJECT *obj, int yes_no); |
void fl_clear_canvas(FL_OBJECT *obj); |
If fl_set_object_color() gas been called on the object
the first color passed to the function will be used to draw the
background of the color, otherwise it's drawn in black.
Some of the attributes, such as boxtype, do not apply to the canvas class.
The first color argument (col1) to
fl_set_object_color() can be used to set the background
color of the canvas (by default, a canvas has no background color).
The second argument (col2) controls the decoration color (if
applicable).
Deriving specialized canvases from the general canvas object is possible. See the next subsection for general approaches how this is done. The following routines work for OpenGL (under X) as well as Mesa, a free OpenGL clone.
To add an OpenGL canvas to a form, use the following routine
FL_OBJECT *fl_add_glcanvas(int type, FL_Coord x, FL_Coord y,
FL_Coord w, FL_Coord h, const char *label);
|
where type is the same as for a normal canvas. A "glcanvas"
created this way will have the following attributes by default
GLX_RGBA, GLX_DEPTH_SIZE: 1, GLX_RED_SIZE: 1, GLX_GREEN_SIZE: 1, GLX_BLUE_SIZE: 1, GLX_DOUBLEBUFFER |
The application program can modify these defaults using the following routine (before the creation of glcanvases)
void fl_set_glcanvas_defaults(const int *attributes); |
See glXChooseVisual() for a list of valid attributes.
To get the current defaults use
void fl_get_glcanvas_defaults(int *attributes); |
It is also possible to change the attributes on a canvas by canvas basis by utilizing the following routine:
void fl_set_glcanvas_attributes(FL_OBJECT *obj, const int *attributes); |
Note that this routine can be used to change a glcanvas attributes on the fly even if the canvas is already visible and active.
To obtain the attributes of a particular canvas, use the following routine
void fl_get_glcanvas_attributes(FL_OBJECT *obj, int attributes[]); |
The caller must supply the space for the attribute values.
To obtain the the glx context (for whatever purposes), use
GLXContext fl_get_glcanvas_context(FL_OBJECT *obj); |
Note that by default the rendering context created by a glcanvas uses direct rendering (i.e., by-passing the Xserver). To change this default, i.e., to always render through the Xserver, use the following routine:
void fl_set_glcanvas_direct(FL_OBJECT *obj, int yes_no); |
with the argument yes_no set to false (0).
Remember that OpenGL drawing routines always draw into the window the current context is bound to. For application with a single canvas, this is not a problem. In case of multiple canvases, the canvas driver takes care of setting the proper context before invoking the expose handler. In some cases, the application may want to draw into canvases actively. In this case, explicit drawing context switching may be required. To this end, use the following routine
void fl_activate_glcanvas(FL_OBJECT *obj); |
before drawing into glcanvas object.
Finally there is a routine that can be used to obtain the XVisual
information that is used to create the context
XVisualInfo *fl_get_glcanvas_xvisualinfo(FL_OBJECT *obj); |
See demo program gl.c for an example use of a glcanvas.
| [ << ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
This document was generated by Peter S Galbraith on February 22, 2014 using texi2html 1.82.