1
0
mirror of git://projects.qi-hardware.com/fped.git synced 2024-11-28 19:09:42 +02:00
fped/inst.c

1424 lines
30 KiB
C
Raw Normal View History

/*
* inst.c - Instance structures
*
* Written 2009-2012, 2015 by Werner Almesberger
* Copyright 2009-2012, 2015 by Werner Almesberger
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*/
#include <stdlib.h>
#include <stdio.h>
#include <math.h>
#include "util.h"
#include "error.h"
#include "coord.h"
#include "expr.h"
#include "layer.h"
#include "obj.h"
#include "delete.h"
#include "gui_util.h"
#include "gui_status.h"
#include "gui_canvas.h"
#include "gui_tool.h"
#include "gui_meas.h"
#include "gui_inst.h"
#include "gui_frame.h"
#include "gui.h"
#include "inst.h"
struct inst *selected_inst = NULL;
struct bbox active_frame_bbox;
struct pkg *pkgs, *active_pkg, *curr_pkg;
struct pkg *reachable_pkg = NULL;
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
struct inst *frame_instantiating = NULL;
static struct pkg *prev_pkgs, *prev_reachable_pkg;
static unsigned long active_set = 0;
static struct inst_ops vec_ops;
static struct inst_ops frame_ops;
static struct inst_ops meas_ops;
#define IS_ACTIVE ((active_set & 1))
/* ----- selective visibility ---------------------------------------------- */
static int show(enum inst_prio prio)
{
switch (prio) {
case ip_vec:
case ip_frame:
return show_stuff;
case ip_meas:
return show_meas;
default:
return 1;
}
}
int bright(const struct inst *inst)
{
if (!show_bright)
return 0;
return inst->ops != &vec_ops && inst->ops != &frame_ops &&
inst->ops != &meas_ops;
}
static int show_this(const struct inst *inst)
{
if (show_all)
return 1;
if (inst->ops == &frame_ops && inst->u.frame.ref == active_frame)
return 1;
if (!inst->outer)
return active_frame == frames;
return inst->outer->u.frame.ref == active_frame;
}
/* ----- selection of items not on the canvas ------------------------------ */
static void *selected_outside = NULL;
static void (*outside_deselect)(void *item);
static void deselect_outside(void)
{
if (selected_outside && outside_deselect)
outside_deselect(selected_outside);
selected_outside = NULL;
}
void inst_select_outside(void *item, void (*deselect)(void *item))
{
if (item == selected_outside)
return;
deselect_outside();
inst_deselect();
selected_outside = item;
outside_deselect = deselect;
}
/* ----- check connectedness ----------------------------------------------- */
/*
* After an instantiation failure, the instances can get out of sync with the
* object tree, and attempts to select an item on the canvas can cause accesses
* to objects that aren't there anymore. So we need to check if we can still
* reach the corresponding object.
*
* Note: even this isn't bullet-proof. Theoretically, we may get a new object
* in the old place. However, this probably doesn't do any serious damage.
*/
static int inst_connected(const struct inst *inst)
{
const struct frame *frame;
const struct vec *vec;
const struct obj *obj;
for (frame = frames; frame; frame = frame->next) {
if (inst->ops == &vec_ops) {
for (vec = frame->vecs; vec; vec = vec->next)
if (vec == inst->vec)
return 1;
} else {
for (obj = frame->objs; obj; obj = obj->next)
if (obj == inst->obj)
return 1;
}
}
return 0;
}
/* ----- selection --------------------------------------------------------- */
static void inst_select_inst(struct inst *inst)
{
selected_inst = inst;
tool_selected_inst(inst);
gui_frame_select_inst(inst);
if (inst->ops->select)
selected_inst->ops->select(inst);
status_set_icon(get_icon_by_inst(inst));
}
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
/*
* @@@ This logic is overly complicated and should be simplified. The general
* idea was to avoid making unnecessary changes to the user's selections, but
* that risk doesn't exist. Furthermore, the way activate_item is used, its
* preconditions aren't met. It works anyway but it could be simpler as a
* consequence.
*
* activate_item tries to activate the path through the frame references,
* leading to a specific instance. It returns whether this is failed or whether
* it may have been successful.
*
* The initial condition is that we want to activate an item on a frame
* instance that's not active. Since the frame has been instantiated, there
* must be a way to activate it. We just have to find out how.
*
* The first test eliminates the root frame. If we're at the root frame and
* still haven't figured out what to do, something is wrong and we give up.
*
* The next test skips references that are already right. Since we know that
* there must be at least one reference that leads elsewhere, and we haven't
* found it yet, the recursion will tell us whether it can find it at all.
*
* Finally, if we've found a mismatch, we correct it. We then try to fix any
* further mismatches. Since we've made progress, we return 1, even if the
* other fixes should fail (or reach the root frame).
*
*/
static int activate_item(struct inst *inst)
{
if (!inst->outer)
return 0;
if (inst->outer->u.frame.ref->active_ref == inst->outer->obj)
return activate_item(inst->outer);
inst->outer->u.frame.ref->active_ref = inst->outer->obj;
activate_item(inst->outer);
return 1;
}
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
static int __inst_select(struct coord pos, int tries)
{
enum inst_prio prio;
const struct inst *prev;
struct inst *inst;
struct inst *first = NULL; /* first active item */
struct inst *next = NULL; /* active item after currently sel. */
struct inst *any_first = NULL; /* first item, active or inactive */
struct inst *any_same_frame = NULL; /* first item on active frame */
struct frame *frame;
int best_dist = 0; /* keep gcc happy */
int select_next;
int dist, i;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
if (!tries) {
fprintf(stderr, "__inst_select: tries exhausted\n");
return 0;
}
prev = selected_inst;
deselect_outside();
edit_nothing();
if (selected_inst) {
gui_frame_deselect_inst(selected_inst);
tool_selected_inst(NULL);
}
inst_deselect();
select_next = 0;
FOR_INST_PRIOS_DOWN(prio) {
if (!show(prio))
continue;
FOR_ALL_INSTS(i, prio, inst) {
if (!show_this(inst))
continue;
if (!inst->ops->distance)
continue;
if (!inst_connected(inst))
continue;
dist = inst->ops->distance(inst, pos, draw_ctx.scale);
if (dist >= 0) {
if (!any_first)
any_first = inst;
if (!any_same_frame && inst->outer &&
inst->outer->u.frame.ref == active_frame)
any_same_frame = inst;
if (!inst->active)
continue;
if (!first)
first = inst;
if (!next && select_next)
next = inst;
if (inst == prev)
select_next = 1;
if (!selected_inst || best_dist > dist) {
selected_inst = inst;
best_dist = dist;
}
}
}
}
if (select_next) {
selected_inst = next ? next : first;
goto selected;
}
if (selected_inst)
goto selected;
/* give vectors a second chance */
if (show_stuff) {
FOR_ALL_INSTS(i, ip_vec, inst) {
if (!inst->active)
continue;
if (!inst_connected(inst))
continue;
dist = gui_dist_vec_fallback(inst, pos, draw_ctx.scale);
if (dist >= 0 && (!selected_inst || best_dist > dist)) {
selected_inst = inst;
best_dist = dist;
}
}
if (selected_inst)
goto selected;
}
if (!show_all)
return 0;
if (any_same_frame) {
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
activate_item(any_same_frame);
search_inst(any_same_frame);
instantiate();
change_world();
return __inst_select(pos, tries-1);
}
if (any_first) {
frame = any_first->outer ? any_first->outer->u.frame.ref : NULL;
if (frame != active_frame) {
select_frame(frame);
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
return __inst_select(pos, tries-1);
}
}
return 0;
selected:
inst_select_inst(selected_inst);
return 1;
}
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
int inst_select(struct coord pos)
{
/*
* We shouldn't need more than 2 tries to select any item, so 5 is more
* than enough. This can still fail, but then it would for any number
* of tries.
*/
return __inst_select(pos, 5);
}
struct inst *inst_find_point(struct coord pos)
{
struct inst *inst, *found;
int best_dist = 0; /* keep gcc happy */
int dist, i;
found = NULL;
FOR_ALL_INSTS(i, ip_frame, inst) {
if (!inst->u.frame.active)
continue;
dist = gui_dist_frame_eye(inst, pos, draw_ctx.scale);
if (dist >= 0 && (!found || best_dist > dist)) {
found = inst;
best_dist = dist;
}
}
if (found)
return found;
FOR_ALL_INSTS(i, ip_vec, inst) {
if (!inst->active || !inst->ops->distance)
continue;
dist = inst->ops->distance(inst, pos, draw_ctx.scale);
if (dist >= 0 && (!found || best_dist > dist)) {
found = inst;
best_dist = dist;
}
}
return found;
}
int inst_find_point_selected(struct coord pos, struct inst **res)
{
struct vec **anchors[3];
int n, best_i, i;
struct inst *best = NULL;
struct inst *inst;
int d_min, d, j;
assert(selected_inst);
n = inst_anchors(selected_inst, anchors);
for (i = 0; i != n; i++) {
if (*anchors[i]) {
FOR_ALL_INSTS(j, ip_vec, inst) {
if (inst->vec != *anchors[i])
continue;
d = gui_dist_vec(inst, pos, draw_ctx.scale);
if (d != -1 && (!best || d < d_min)) {
best = inst;
best_i = i;
d_min = d;
}
}
} else {
FOR_ALL_INSTS(j, ip_frame, inst) {
if (inst != selected_inst->outer)
continue;
d = gui_dist_frame(inst, pos, draw_ctx.scale);
if (d != -1 && (!best || d < d_min)) {
best = inst;
best_i = i;
d_min = d;
}
}
}
}
if (!best)
return -1;
if (res)
*res = best;
return best_i;
}
struct coord inst_get_point(const struct inst *inst)
{
if (inst->ops == &vec_ops)
return inst->u.vec.end;
if (inst->ops == &frame_ops)
return inst->base;
abort();
}
struct vec *inst_get_vec(const struct inst *inst)
{
if (inst->ops == &vec_ops)
return inst->vec;
if (inst->ops == &frame_ops)
return NULL;
abort();
}
int inst_anchors(struct inst *inst, struct vec ***anchors)
{
if (inst->vec) {
anchors[0] = &inst->vec->base;
return 1;
}
return obj_anchors(inst->obj, anchors);
}
void inst_deselect(void)
{
if (selected_inst) {
tool_selected_inst(NULL);
gui_frame_deselect_inst(selected_inst);
}
deselect_outside();
status_set_type_x(NULL, "");
status_set_type_y(NULL, "");
status_set_type_entry(NULL, "");
status_set_name(NULL, "");
status_set_x(NULL, "");
status_set_y(NULL, "");
status_set_r(NULL, "");
status_set_angle(NULL, "");
selected_inst = NULL;
edit_nothing();
refresh_pos();
status_set_icon(NULL);
}
/* ----- select instance by vector/object ---------------------------------- */
static void vec_edit(struct vec *vec);
static void obj_edit(struct obj *obj);
void inst_select_vec(struct vec *vec)
{
struct inst *inst;
int i;
if (vec->frame != active_frame)
select_frame(vec->frame);
FOR_ALL_INSTS(i, ip_vec, inst)
if (inst->vec == vec && inst->active) {
inst_deselect();
inst_select_inst(inst);
return;
}
vec_edit(vec);
}
void inst_select_obj(struct obj *obj)
{
enum inst_prio prio;
struct inst *inst;
int i;
if (obj->frame != active_frame)
select_frame(obj->frame);
FOR_INST_PRIOS_DOWN(prio)
FOR_ALL_INSTS(i, prio, inst)
if (inst->obj && inst->obj == obj && inst->active)
goto found;
obj_edit(obj);
return;
found:
inst_deselect();
inst_select_inst(inst);
}
/* ----- common status reporting ------------------------------------------- */
static void rect_status(struct coord a, struct coord b, unit_type width,
int rounded)
{
const char *tip;
struct coord d = sub_vec(b, a);
double r;
unit_type diag;
status_set_xy(d);
tip = "Angle of diagonal";
if (!d.x && !d.y) {
status_set_angle(tip, "a = 0 deg");
} else {
status_set_angle(tip, "a = %3.1f deg", theta(a, b));
}
if (d.x < 0)
d.x = -d.x;
if (d.y < 0)
d.y = -d.y;
diag = hypot(d.x, d.y);
if (rounded) {
/*
* Only consider the part of the diagonal that is on the pad
* surface.
*
* The circle: (x-r)^2+(y-r)^2 = r^2
* The diagonal: x = t*cos(theta), y = t*sin(theta)
*
* t is the distance from the corner of the surrounding
* rectangle to the half-circle:
*
* t = 2*r*(s+c-sqrt(2*s*c))
*
* With s = sin(theta) and c = cos(theta).
*
* Since d.x = diag*cos(theta), we don't need to calculate the
* sinus and cosinus but can use d.x and d.y directly.
*/
r = (d.x > d.y ? d.y : d.x)/2;
diag -= 2*r*(d.x+d.y-sqrt(2*d.x*d.y))/diag;
}
set_with_units(status_set_r, "d = ", diag, "Length of diagonal");
if (width != -1) {
status_set_type_entry(NULL, "width =");
set_with_units(status_set_name, "", width, "Line width");
}
}
static void rect_status_sort(struct coord a, struct coord b, unit_type width,
int rounded)
{
sort_coord(&a, &b);
rect_status(a, b, width, rounded);
}
/* ----- helper functions for instance creation ---------------------------- */
static void update_bbox(struct bbox *bbox, struct coord coord)
{
if (bbox->min.x > coord.x)
bbox->min.x = coord.x;
if (bbox->max.x < coord.x)
bbox->max.x = coord.x;
if (bbox->min.y > coord.y)
bbox->min.y = coord.y;
if (bbox->max.y < coord.y)
bbox->max.y = coord.y;
}
static void propagate_bbox(const struct inst *inst)
{
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
struct inst *frame = frame_instantiating ?
frame_instantiating : curr_pkg->insts[ip_frame];
update_bbox(&frame->bbox, inst->bbox.min);
update_bbox(&frame->bbox, inst->bbox.max);
if (curr_pkg->bbox.min.x || curr_pkg->bbox.min.y ||
curr_pkg->bbox.max.x || curr_pkg->bbox.max.y) {
update_bbox(&curr_pkg->bbox, inst->bbox.min);
update_bbox(&curr_pkg->bbox, inst->bbox.max);
} else {
curr_pkg->bbox = inst->bbox;
}
}
static void grow_bbox_by_width(struct bbox *bbox, unit_type width)
{
bbox->min.x -= width/2;
bbox->min.y -= width/2;
bbox->max.x += width/2;
bbox->max.y += width/2;
}
static int zero_sized(struct coord a, struct coord b, const char *fmt,
const char *arg)
{
if (a.x == b.x && a.y == b.y) {
fail(fmt, "zero-sized", arg);
return 1;
}
if (a.x == b.x) {
fail(fmt, "zero-width", arg);
return 1;
}
if (a.y == b.y) {
fail(fmt, "zero-height", arg);
return 1;
}
return 0;
}
static struct inst *add_inst(const struct inst_ops *ops, enum inst_prio prio,
struct coord base)
{
struct inst *inst;
inst = alloc_type(struct inst);
inst->ops = ops;
inst->prio = prio;
inst->vec = NULL;
inst->obj = NULL;
inst->base = inst->bbox.min = inst->bbox.max = base;
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
inst->outer = frame_instantiating;
inst->active = IS_ACTIVE;
inst->next = NULL;
*curr_pkg->next_inst[prio] = inst;
curr_pkg->next_inst[prio] = &inst->next;
return inst;
}
/* ----- vec --------------------------------------------------------------- */
static int validate_vec_name(const char *s, void *ctx)
{
struct vec *vec = ctx;
const struct vec *walk;
if (!is_id(s))
return 0;
for (walk = vec->frame->vecs; walk; walk = walk->next)
if (walk->name && !strcmp(walk->name, s))
return 0;
return 1;
}
static void vec_edit(struct vec *vec)
{
edit_x(&vec->x, "X distance");
edit_y(&vec->y, "Y distance");
edit_unique_null(&vec->name, validate_vec_name, vec, "Vector name");
}
static void vec_op_select(struct inst *self)
{
status_set_type_entry(NULL, "ref =");
status_set_name("Vector reference (name)",
"%s", self->vec->name ? self->vec->name : "");
rect_status(self->base, self->u.vec.end, -1, 0);
vec_edit(self->vec);
}
/*
* @@@ The logic of gui_find_point_vec isn't great. Instead of selecting a
* point and then filtering, we should filter the candidates, so that a point
* that's close end eligible can win against one that's closer but not
* eligible.
*/
static struct inst *find_point_vec(struct inst *self, struct coord pos)
{
struct inst *inst;
const struct vec *vec;
inst = inst_find_point(pos);
if (!inst)
return NULL;
if (inst->ops == &frame_ops)
return inst;
for (vec = inst->vec; vec; vec = vec->base)
if (vec == self->vec)
return NULL;
return inst;
}
/*
* When instantiating and when dumping, we assume that bases appear in the
* frame->vecs list before vectors using them. A move may change this order.
* We therefore have to sort the list after the move.
*
* Since the list is already ordered, cleaning it up is just O(n).
*/
static void do_move_to_vec(struct inst *inst, struct inst *to, int i)
{
struct vec *to_vec = inst_get_vec(to);
struct vec *vec = inst->vec;
struct frame *frame = vec->frame;
struct vec *v, **anchor, **walk;
assert(!i);
vec->base = to_vec;
/*
* Mark the vector that's being rebased and all vectors that
* (recursively) depend on it.
*
* We're mainly interested in the range between the vector being moved
* and the new base. If the vector follows the base, the list is
* already in the correct order and nothing needs moving.
*/
for (v = frame->vecs; v != vec; v = v->next)
v->mark = 0;
vec->mark = 1;
for (v = vec->next; v && v != to_vec; v = v->next)
v->mark = v->base ? v->base->mark : 0;
if (!v)
return;
/*
* All the marked vectors appearing on the list before the new base
* are moved after the new base, preserving their order.
*
* Start at frame->vecs, not "vec", so that we move the the vector
* being rebased as well.
*/
anchor = &to_vec->next;
walk = &frame->vecs;
while (*walk != to_vec) {
v = *walk;
if (!v->mark) {
walk = &v->next;
} else {
*walk = v->next;
v->next = *anchor;
*anchor = v;
anchor = &v->next;
}
}
}
static struct inst_ops vec_ops = {
.draw = gui_draw_vec,
.hover = gui_hover_vec,
.distance = gui_dist_vec,
.select = vec_op_select,
.find_point = find_point_vec,
.draw_move = draw_move_vec,
.do_move_to = do_move_to_vec,
};
int inst_vec(struct vec *vec, struct coord base)
{
struct inst *inst;
inst = add_inst(&vec_ops, ip_vec, base);
inst->vec = vec;
inst->u.vec.end = vec->pos;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
update_bbox(&inst->bbox, vec->pos);
propagate_bbox(inst);
return 1;
}
/* ----- line -------------------------------------------------------------- */
static void obj_line_edit(struct obj *obj)
{
edit_dist_expr(&obj->u.line.width, "Line width");
}
static void line_op_select(struct inst *self)
{
rect_status_sort(self->base, self->u.rect.end, self->u.rect.width, 0);
obj_line_edit(self->obj);
}
static struct inst_ops line_ops = {
.draw = gui_draw_line,
.distance = gui_dist_line,
.select = line_op_select,
.draw_move = draw_move_line,
};
int inst_line(struct obj *obj, struct coord a, struct coord b, unit_type width)
{
struct inst *inst;
inst = add_inst(&line_ops, ip_line, a);
inst->obj = obj;
inst->u.rect.end = b;
inst->u.rect.width = width;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
update_bbox(&inst->bbox, b);
grow_bbox_by_width(&inst->bbox, width);
propagate_bbox(inst);
return 1;
}
/* ----- rect -------------------------------------------------------------- */
static void obj_rect_edit(struct obj *obj)
{
edit_dist_expr(&obj->u.rect.width, "Line width");
}
static void rect_op_select(struct inst *self)
{
rect_status_sort(self->base, self->u.rect.end, self->u.rect.width, 0);
obj_rect_edit(self->obj);
}
static struct inst_ops rect_ops = {
.draw = gui_draw_rect,
.distance = gui_dist_rect,
.select = rect_op_select,
.draw_move = draw_move_rect,
};
int inst_rect(struct obj *obj, struct coord a, struct coord b, unit_type width)
{
struct inst *inst;
inst = add_inst(&rect_ops, ip_rect, a);
inst->obj = obj;
inst->u.rect.end = b;
inst->u.rect.width = width;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
update_bbox(&inst->bbox, b);
grow_bbox_by_width(&inst->bbox, width);
propagate_bbox(inst);
return 1;
}
/* ----- pad / rpad -------------------------------------------------------- */
static int validate_pad_name(const char *s, void *ctx)
{
char *tmp;
status_begin_reporting();
tmp = expand(s, NULL);
if (!tmp)
return 0;
free(tmp);
return 1;
}
static void obj_pad_edit(struct obj *obj)
{
edit_pad_type(&obj->u.pad.type);
edit_name(&obj->u.pad.name, validate_pad_name, NULL,
"Pad name (template)");
}
static void pad_op_select(struct inst *self)
{
status_set_type_entry(NULL, "label =");
status_set_name("Pad name (actual)", "%s", self->u.pad.name);
rect_status_sort(self->base, self->u.pad.other, -1, 0);
obj_pad_edit(self->obj);
}
static struct inst_ops pad_ops = {
.draw = gui_draw_pad,
.distance = gui_dist_pad,
.select = pad_op_select,
.draw_move = draw_move_pad,
};
static void rpad_op_select(struct inst *self)
{
status_set_type_entry(NULL, "label =");
status_set_name("Pad name (actual)", "%s", self->u.pad.name);
rect_status_sort(self->base, self->u.pad.other, -1, 1);
obj_pad_edit(self->obj);
}
static struct inst_ops rpad_ops = {
.draw = gui_draw_rpad,
.distance = gui_dist_pad, /* @@@ */
.select = rpad_op_select,
.draw_move = draw_move_rpad,
};
int inst_pad(struct obj *obj, const char *name, struct coord a, struct coord b)
{
struct inst *inst;
if (zero_sized(a, b, "%s pad \"%s\"", name))
return 0;
inst = add_inst(obj->u.pad.rounded ? &rpad_ops : &pad_ops,
obj->u.pad.type == pt_normal || obj->u.pad.type == pt_bare ||
obj->u.pad.type == pt_trace ?
ip_pad_copper : ip_pad_special, a);
inst->obj = obj;
inst->u.pad.name = stralloc(name);
inst->u.pad.other = b;
inst->u.pad.layers = pad_type_to_layers(obj->u.pad.type);
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
update_bbox(&inst->bbox, b);
propagate_bbox(inst);
return 1;
}
/* ----- hole -------------------------------------------------------------- */
static void hole_op_select(struct inst *self)
{
rect_status_sort(self->base, self->u.hole.other, -1, 1);
}
static struct inst_ops hole_ops = {
.draw = gui_draw_hole,
.distance = gui_dist_hole,
.select = hole_op_select,
.draw_move = draw_move_hole,
};
int inst_hole(struct obj *obj, struct coord a, struct coord b)
{
struct inst *inst;
if (zero_sized(a, b, "%s hole", NULL))
return 0;
inst = add_inst(&hole_ops, ip_hole, a);
inst->obj = obj;
inst->u.hole.other = b;
inst->u.hole.layers = mech_hole_layers();
find_inst(inst);
update_bbox(&inst->bbox, b);
propagate_bbox(inst);
return 1;
}
/* ----- arc --------------------------------------------------------------- */
static void obj_arc_edit(struct obj *obj)
{
edit_dist_expr(&obj->u.arc.width, "Line width");
}
static void arc_op_select(struct inst *self)
{
status_set_xy(self->base);
status_set_angle("Angle", "a = %3.1f deg",
self->u.arc.a1 == self->u.arc.a2 ? 360 :
self->u.arc.a2-self->u.arc.a1);
set_with_units(status_set_r, "r = ", self->u.arc.r, "Radius");
status_set_type_entry(NULL, "width =");
set_with_units(status_set_name, "", self->u.arc.width, "Line width");
obj_arc_edit(self->obj);
}
static struct inst_ops arc_ops = {
.draw = gui_draw_arc,
.distance = gui_dist_arc,
.select = arc_op_select,
.draw_move = draw_move_arc,
.do_move_to = do_move_to_arc,
};
int inst_arc(struct obj *obj, struct coord center, struct coord start,
struct coord end, unit_type width)
{
struct inst *inst;
double r, a1, a2;
a1 = theta(center, start);
a2 = theta(center, end);
inst = add_inst(&arc_ops,
fmod(a1, 360) == fmod(a2, 360) ? ip_circ : ip_arc, center);
inst->obj = obj;
r = hypot(start.x-center.x, start.y-center.y);
inst->u.arc.r = r;
inst->u.arc.a1 = a1;
inst->u.arc.a2 = a2;
inst->u.arc.width = width;
inst->bbox.min.x = center.x-r;
inst->bbox.max.x = center.x+r;
inst->bbox.min.y = center.y-r;
inst->bbox.max.y = center.y+r;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
grow_bbox_by_width(&inst->bbox, width);
propagate_bbox(inst);
return 1;
}
/* ----- measurement ------------------------------------------------------- */
static void obj_meas_edit(struct obj *obj)
{
edit_dist_expr(&obj->u.meas.offset, "Measurement line offset");
}
static void meas_op_select(struct inst *self)
{
rect_status_sort(self->base, self->u.meas.end, -1, 0);
status_set_type_entry(NULL, "offset =");
set_with_units(status_set_name, "", self->u.meas.offset,
"Measurement line offset");
obj_meas_edit(self->obj);
}
static struct inst_ops meas_ops = {
.draw = gui_draw_meas,
.distance = gui_dist_meas,
.select = meas_op_select,
.begin_drag_move= begin_drag_move_meas,
.find_point = find_point_meas_move,
.draw_move = draw_move_meas,
.end_drag_move = end_drag_move_meas,
.do_move_to = do_move_to_meas,
};
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
struct inst *find_meas_hint(const struct obj *obj)
{
struct inst *inst;
for (inst = curr_pkg->insts[ip_meas]; inst; inst = inst->next)
if (inst->obj == obj)
break;
return inst;
}
int inst_meas(struct obj *obj, struct coord from, struct coord to)
{
struct inst *inst;
struct coord a1, b1;
inst = find_meas_hint(obj);
assert(inst);
inst->base = from;
inst->u.meas.end = to;
inst->u.meas.valid = 1;
/* @@@ we still need to consider the text size as well */
update_bbox(&inst->bbox, from);
update_bbox(&inst->bbox, to);
project_meas(inst, &a1, &b1);
update_bbox(&inst->bbox, a1);
update_bbox(&inst->bbox, b1);
propagate_bbox(inst);
return 1;
}
void inst_meas_hint(struct obj *obj, unit_type offset)
{
static const struct coord zero = { 0, 0 };
struct inst *inst;
inst = find_meas_hint(obj);
if (inst)
return;
inst = add_inst(&meas_ops, ip_meas, zero);
inst->obj = obj;
inst->u.meas.offset = offset;
inst->u.meas.valid = 0;
inst->active = 1; /* measurements are always active */
}
/* ----- direct editing of objects ----------------------------------------- */
static void obj_edit(struct obj *obj)
{
switch (obj->type) {
case ot_frame:
break;
case ot_line:
obj_line_edit(obj);
break;
case ot_rect:
obj_rect_edit(obj);
break;
case ot_arc:
obj_arc_edit(obj);
break;
case ot_pad:
obj_pad_edit(obj);
break;
case ot_meas:
obj_meas_edit(obj);
break;
default:
abort();
}
}
/* ----- active instance --------------------------------------------------- */
void inst_begin_active(int active)
{
active_set = (active_set << 1) | active;
}
void inst_end_active(void)
{
active_set >>= 1;
}
/* ----- frame ------------------------------------------------------------- */
static void frame_op_select(struct inst *self)
{
rect_status(self->bbox.min, self->bbox.max, -1, 0);
status_set_type_entry(NULL, "name =");
status_set_name("Frame name", "%s", self->u.frame.ref->name);
}
static struct inst_ops frame_ops = {
.draw = gui_draw_frame,
.hover = gui_hover_frame,
.distance = gui_dist_frame,
.select = frame_op_select,
.draw_move = draw_move_frame,
};
void inst_begin_frame(struct obj *obj, struct frame *frame,
struct coord base, int active, int is_active_frame)
{
struct inst *inst;
inst = add_inst(&frame_ops, ip_frame, base);
inst->obj = obj;
inst->u.frame.ref = frame;
inst->u.frame.active = is_active_frame;
inst->active = active;
When clicking on an instance, fped used to select the currenly active instance of the corresponding object but didn't change any table or loop selection to make it active. (It did already change the frame reference.) This meant that, upon clicking on an instance, often a different instance of the same object would be selected. This was confusing at best. With the below changes, table rows and loop iterations are adjusted such that the instance that was clicked on becomes active. If the algorithm fails, fped will print "__inst_select: tries exhausted" and fail safely. Please report if this happens. - obj.c (search_inst, find_inst, instantiate): added mechanism to search for instances matching a previous instance - obj.c (run_loops, iterate_tables): record matches in found_* elements of the object's struct - obj.c (reset_found, activate_found): helper functions to initialize and apply the activation leading to the instance found - inst.c (activate_item): added comment explaining how activate_item is supposed to work and the fallacies of that logic - inst.c (inst_select): added tries counter to avoid infinite recursion when results don't converge - inst.c (__inst_select): when selecting an instance in the same frame, adjust tables and loops such that the instance becomes active - inst.c: added call to find_inst after most instance creations (add_inst) - obj.h: documented the meaning of the curr[ent]*, active*, and found* fields git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5792 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-01-12 04:18:58 +02:00
find_inst(inst);
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
frame_instantiating = inst;
}
void inst_end_frame(const struct frame *frame)
{
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
struct inst *inst = frame_instantiating;
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
frame_instantiating = frame_instantiating->outer;
if (frame_instantiating)
propagate_bbox(inst);
if (inst->u.frame.active && frame == active_frame)
active_frame_bbox = inst->bbox;
}
/* ----- package ----------------------------------------------------------- */
void inst_select_pkg(const char *name, int active)
{
struct pkg **pkg;
enum inst_prio prio;
name = name ? unique(name) : NULL;
for (pkg = &pkgs; *pkg; pkg = &(*pkg)->next)
if ((*pkg)->name == name)
break;
if (!*pkg) {
*pkg = zalloc_type(struct pkg);
(*pkg)->name = name;
FOR_INST_PRIOS_UP(prio)
(*pkg)->next_inst[prio] = &(*pkg)->insts[prio];
(*pkg)->samples =
zalloc_size(sizeof(struct sample *)*n_samples);
(*pkg)->n_samples = n_samples;
}
curr_pkg = *pkg;
if (active && name)
reachable_pkg = curr_pkg;
}
/* ----- misc. ------------------------------------------------------------- */
struct bbox inst_get_bbox(const struct pkg *pkg)
{
if (pkg)
return pkg->bbox;
else
return pkgs->insts[ip_frame]->bbox;
}
static void cleanup_inst(enum inst_prio prio, const struct inst *inst)
{
switch (prio) {
case ip_pad_copper:
case ip_pad_special:
free(inst->u.pad.name);
break;
default:
break;
}
}
static void free_pkgs(struct pkg *pkg)
{
enum inst_prio prio;
struct pkg *next_pkg;
struct inst *inst, *next;
while (pkg) {
next_pkg = pkg->next;
FOR_INST_PRIOS_UP(prio)
for (inst = pkg->insts[prio]; inst; inst = next) {
next = inst->next;
cleanup_inst(prio, inst);
free(inst);
}
reset_samples(pkg->samples, pkg->n_samples);
free(pkg->samples);
free(pkg);
pkg = next_pkg;
}
}
void inst_start(void)
{
static struct bbox bbox_zero = { { 0, 0 }, { 0, 0 }};
active_frame_bbox = bbox_zero;
prev_pkgs = pkgs;
prev_reachable_pkg = reachable_pkg;
pkgs = NULL;
reachable_pkg = NULL;
inst_select_pkg(NULL, 0);
curr_pkg = pkgs;
The mechanism for selecting points for measurements reaches its limits when using frames to encapsulate building blocks, e.g., like macros or functions in a programming language. Since measurements only know about the frame containing a vector but not the frames containing that frame, invocations of this frame from different places can only be distinguished within the min/next/max scheme. (See the example in README.) To eliminate this limitation, one needs a way to tell fped to consider a point only if it has been instantiated through a certain path, e.g., by requiring some other frames to be visited in its instantiation. This increases the number of distinct points available for measurements. The mechanism chosen is to qualify a measurement point with frames that lead to it. This list of outer frames does not have to include all frames. Without qualifying, the old behaviour results. Note that this doesn't cover all possible ways in which a point can appear in different roles. Multiple frame references can also result from repeating the same frame reference in the same parent frame. The current qualification mechanism does not allow such paths to be distinguished. However, one can always introduce intermediate frames for this purpose. Furthermore, repetitions create multiple instances of a point, although in what should be considered the same role. - fpd.l: make scanner support free-format a little better by switching back to keyword mode after frame braces. This way, one can write a simple frame in a single line, which is useful for regression tests. - fpd.l, fpd.y, README, test/dbg_meas: added %meas directive to print the result of a measurement - fpd.y, README: measurements can now be labeled. Note that, due to limitations of the grammar, the first measurement cannot be labeled. - error.h, error.c (yywarn): new function for non-fatal diagnostics that always get reported to standard error - bitset.h, bitset.c: functions to manipulate variable-size bit sets - meas.h, fpd.y, README, test/meas_qual: added the means to specify qualifiers for points used in measurements - dump.c (print_meas_base, print_meas): dump qualifiers - delete.c (delete_references, test/del_frame): delete measurements that reference a frame being deleted in their qualifiers - obj.h, obj.c (enumerate_frames, instantiate): enumerate all frames so that we have an index into the bit vector of visited frames - meas.h, meas.c (reset_samples, meas_post), obj.c (generate_vecs, generate_frame, instantiate): record the set of frames visited for each sample - meas.c (meas_post): only treat two instances of a point as equivalent if the set of frames visited of one of them is a superset of set of the other. In this case, keep the union of the two sets. - meas.h, meas.c (meas_find_min, meas_find_next, meas_find_max), test/meas_qual: instantiate_meas_pkg only select points for which all frames in the qualification have been visited - gui_meas.c (is_min, is_next, is_max, is_a_next): updated for above change - inst.h, inst.c (curr_frame, propagate_bbox, add_inst, inst_begin_frame, inst_end_frame, inst_start): renamed curr_frame to frame_instantiating to avoid clash with curr_frame in fpd.y - inst.h, inst.c (find_meas_hint): make global - test/structure, test/del_vec, test/del_frame: fped now warns if a measurement is in an unlinked frame. Changed regressions tests to avoid this warning. - test/Common: new function expect_grep to compare only part of the output git-svn-id: http://svn.openmoko.org/trunk/eda/fped@5967 99fdad57-331a-0410-800a-d7fa5415bdb3
2010-05-30 00:13:48 +03:00
frame_instantiating = NULL;
}
void inst_commit(void)
{
struct pkg *pkg;
if (active_pkg) {
for (pkg = pkgs; pkg && pkg->name != active_pkg->name;
pkg = pkg->next);
active_pkg = pkg;
}
if (!active_pkg)
active_pkg = pkgs->next;
free_pkgs(prev_pkgs);
}
void inst_revert(void)
{
free_pkgs(pkgs);
pkgs = prev_pkgs;
reachable_pkg = prev_reachable_pkg;
}
void inst_draw(void)
{
enum inst_prio prio;
struct inst *inst;
int i;
FOR_INST_PRIOS_UP(prio)
FOR_ALL_INSTS(i, prio, inst)
if (show_this(inst))
if (show(prio) && !inst->active &&
inst->ops->draw)
inst->ops->draw(inst);
FOR_INST_PRIOS_UP(prio)
FOR_ALL_INSTS(i, prio, inst)
if (show(prio) && prio != ip_frame && inst->active &&
inst != selected_inst && inst->ops->draw)
inst->ops->draw(inst);
if (show_stuff)
FOR_ALL_INSTS(i, ip_frame, inst)
if (inst->active && inst != selected_inst &&
inst->ops->draw)
inst->ops->draw(inst);
if (selected_inst && selected_inst->ops->draw)
selected_inst->ops->draw(selected_inst);
}
void inst_highlight_vecs(int (*pick)(struct inst *inst, void *user), void *user)
{
struct inst *inst;
int i;
FOR_ALL_INSTS(i, ip_vec, inst) {
inst->u.vec.highlighted = pick(inst, user);
if (inst->u.vec.highlighted)
gui_highlight_vec(inst);
}
}
struct inst *inst_find_vec(struct coord pos,
int (*pick)(struct inst *inst, void *user), void *user)
{
struct inst *inst, *found;
int best_dist = 0; /* keep gcc happy */
int dist, i;
found = NULL;
FOR_ALL_INSTS(i, ip_vec, inst) {
if (!inst->ops->distance)
continue;
dist = inst->ops->distance(inst, pos, draw_ctx.scale);
if (dist < 0 || (found && best_dist <= dist))
continue;
if (!pick(inst, user))
continue;
found = inst;
best_dist = dist;
}
return found;
}
struct inst *insts_ip_vec(void)
{
return active_pkg->insts[ip_vec];
}
struct pix_buf *inst_draw_move(struct inst *inst, struct coord pos, int i)
{
return inst->ops->draw_move(inst, pos, i);
}
int inst_do_move_to(struct inst *inst, struct inst *to, int i)
{
if (!inst->ops->do_move_to)
return 0;
inst->ops->do_move_to(inst, to, i);
return 1;
}
struct pix_buf *inst_hover(struct inst *inst)
{
if (!inst->ops->hover)
return NULL;
return inst->ops->hover(inst);
}
void inst_begin_drag_move(struct inst *inst, int i)
{
if (inst->ops->begin_drag_move)
inst->ops->begin_drag_move(inst, i);
}
void inst_delete(struct inst *inst)
{
if (inst->ops == &vec_ops)
delete_vec(inst->vec);
else
delete_obj(inst->obj);
}