sac2c issueshttps://gitlab.sac-home.org/sac-group/sac2c/-/issues2017-11-19T21:53:45Zhttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/2059objects inhibit specialization directive2017-11-19T21:53:45ZSven-Bodo Scholzobjects inhibit specialization directive| | |
| --- | --- |
| Bugzilla Link | [429](http://bugs.sac-home.org/show_bug.cgi?id=429) |
| Created on | May 30, 2008 14:19 |
| Resolution | FIXED |
| Resolved on | Jun 01, 2008 05:22 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [429](http://bugs.sac-home.org/show_bug.cgi?id=429) |
| Created on | May 30, 2008 14:19 |
| Resolution | FIXED |
| Resolved on | Jun 01, 2008 05:22 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [tutu.sac](/uploads/efb255199cad3da8fb1d4bbecea1aed5/tutu.sac) |
## Extended Description
<pre>Created an attachment (id=472)
source code
It is impossible to enforce sprecializations for functions that make use of objects. What happens is that the object transformation does not affect the specializations and therefore the specializer thinks that these are different (non overloaded) functions!
rev 15709 of the compiler gives the following on error message for the attached program:
ABORT: line 2 file: tutu.sac
ABORT: No definition found for a function "_MAIN::foo" that accepts an
ABORT: argument of type "int" as parameter no 1. Full argument types are "(
ABORT: int)".</pre>Stephan HerhutStephan Herhuthttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1698prop_obj prfs missing2017-11-19T21:16:53ZKai Trojahnerprop_obj prfs missing| | |
| --- | --- |
| Bugzilla Link | [311](http://bugs.sac-home.org/show_bug.cgi?id=311) |
| Created on | Oct 23, 2006 15:53 |
| Resolution | FIXED |
| Resolved on | Nov 09, 2007 12:35 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [311](http://bugs.sac-home.org/show_bug.cgi?id=311) |
| Created on | Oct 23, 2006 15:53 |
| Resolution | FIXED |
| Resolved on | Nov 09, 2007 12:35 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [wldpbug.sac](/uploads/205b482d72bb7364b048634a95e8dc2a/wldpbug.sac) |
## Extended Description
<pre>Compiling a stepped mandelbrot example fails
int mandelbrot( double xmin, double xmax, double ymin, int depth, double zoom,
SDLdisplay &disp )
{
drawn = genarray( [YRES, XRES], false );
steps = [16,4];
for( i=0; i < 2; i++ )
{
drawn, disp = with
{
( . <= [y,x] <= . step [steps[[i]], steps[[i]]] ): ( true, disp );
} ( genarray( [YRES, XRES], false ) , propagate( disp ) );
}
It turns out, that in Withloop-default partition generation (-b10:wldp, after
-b10:accu), the important prop_obj prfs are not compiled and thus we get faulty
reference counting on disp.
Since we are not really interesting in locking the display in these generators,
I suggest a comeback of the F_prop_obj. F_prop_obj could be the result of
merging F_prop_obj_in, F_prop_obj_out and it can be used in cases like this
where we are only interested in a correct representation.</pre>Theo van KlaverenTheo van Klaverenhttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1697prop_obj_out is not eliminated in with-loop unrolling2017-11-19T21:16:47ZRobert Berneckyprop_obj_out is not eliminated in with-loop unrolling| | |
| --- | --- |
| Bugzilla Link | [304](http://bugs.sac-home.org/show_bug.cgi?id=304) |
| Created on | Oct 11, 2006 23:37 |
| Resolution | FIXED |
| Resolved on | Jun 12, 2007 17:54 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [304](http://bugs.sac-home.org/show_bug.cgi?id=304) |
| Created on | Oct 11, 2006 23:37 |
| Resolution | FIXED |
| Resolved on | Jun 12, 2007 17:54 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [nmo.sac](/uploads/7d321c6efc302a46b79f6e8262314064/nmo.sac), [xnmo.sac](/uploads/377d9dd7cfc760e80a35c25d4b24c79c/xnmo.sac) |
## Extended Description
<pre>This is build # 15025
If I compile with -DBUG, as shown, I get this:
ABORT: line 870 file: nmo.sac
ABORT: No definition found for a function "plusDDD" that accepts an argument
ABORT: of type "double" as parameter no 1
*** Compilation failed ***
*** Exit code 6 (Running type inference system)
*** 1 Error(s), 64 Warning(s)
The relevant bit of code is this:
inline double plusslXDD(double[.] y)
{ /* First/last axis fold-based reduction of vector */
lim = shape(y)[0]-1;
z = with (0*shape(y) <= iv < shape(y))
#ifdef BUG
fold( plusDDD, toD(0), toD(y[lim-iv]));
#else
fold( +, toD(0), toD(y[lim-iv]));
#endif
return(z);
}
inline double plusDDD(double x, double y)
{ return(x+y);
}
apex@rattler:~/sandbox> sac2c -DBUG nmo.sac
If I compile with:
sac2c -check b -maxwlur 3 -O3 nmo.sac
I get a crash in phase 15, "marking memval identifiers"</pre>Theo van KlaverenTheo van Klaverenhttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1691void-with-loop and object2017-11-19T21:16:10ZTorben Gerhards void-with-loop and object| | |
| --- | --- |
| Bugzilla Link | [373](http://bugs.sac-home.org/show_bug.cgi?id=373) |
| Created on | Jun 13, 2007 10:18 |
| Resolution | FIXED |
| Resolved on | Jun 14, 2007 11:24 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [373](http://bugs.sac-home.org/show_bug.cgi?id=373) |
| Created on | Jun 13, 2007 10:18 |
| Resolution | FIXED |
| Resolved on | Jun 14, 2007 11:24 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [Fifo.sac](/uploads/fa9b85ed2f57f30973e747d7fa947ff5/Fifo.sac), [fifo_with_bug.sac](/uploads/65c801f27af152e1a2105a131d10d678/fifo_with_bug.sac) |
## Extended Description
<pre>if increasing an array-field of an object inside a void with-loop, another
untouched field contains strange values (afterwards).</pre>Torben Gerhards Torben Gerhards https://gitlab.sac-home.org/sac-group/sac2c/-/issues/1620Mandelbrot crashes if function mandelbrot is not inlined2017-11-19T21:08:48ZKai TrojahnerMandelbrot crashes if function mandelbrot is not inlined| | |
| --- | --- |
| Bugzilla Link | [296](http://bugs.sac-home.org/show_bug.cgi?id=296) |
| Created on | Sep 28, 2006 14:46 |
| Resolution | INVALID |
| Resolved on | Oct 24, 2006 11:50 |
| Version | 1.00beta |
| OS | Linux |
| Archi...| | |
| --- | --- |
| Bugzilla Link | [296](http://bugs.sac-home.org/show_bug.cgi?id=296) |
| Created on | Sep 28, 2006 14:46 |
| Resolution | INVALID |
| Resolved on | Oct 24, 2006 11:50 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [mandelbrot_opt.sac](/uploads/cdc0cd20034ae43a7c6ea1b6a955cf52/mandelbrot_opt.sac) |
## Extended Description
<pre>Very strange behaviour...
Look at the attached example, compile it with -DWORKS to see a running and
without the option to see it crashing.</pre>Stephan HerhutStephan Herhuthttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1613sac2c crashes during object restoration2017-11-19T21:08:03ZKai Trojahnersac2c crashes during object restoration| | |
| --- | --- |
| Bugzilla Link | [222](http://bugs.sac-home.org/show_bug.cgi?id=222) |
| Created on | May 18, 2006 16:25 |
| Resolution | FIXED |
| Resolved on | May 19, 2006 12:27 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [222](http://bugs.sac-home.org/show_bug.cgi?id=222) |
| Created on | May 18, 2006 16:25 |
| Resolution | FIXED |
| Resolved on | May 19, 2006 12:27 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [bug203.sac](/uploads/69c1f06f2c05970f807b989706d58255/bug203.sac) |
## Extended Description
summary says it allStephan HerhutStephan Herhuthttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1427Additional parameter BEFORE reference arg kill sac2c in phase 182017-11-19T20:49:33ZKai TrojahnerAdditional parameter BEFORE reference arg kill sac2c in phase 18| | |
| --- | --- |
| Bugzilla Link | [300](http://bugs.sac-home.org/show_bug.cgi?id=300) |
| Created on | Oct 09, 2006 17:22 |
| Resolution | FIXED |
| Resolved on | Oct 09, 2006 18:26 |
| Version | 1.00beta |
| OS | Linux |
| Archite...| | |
| --- | --- |
| Bugzilla Link | [300](http://bugs.sac-home.org/show_bug.cgi?id=300) |
| Created on | Oct 09, 2006 17:22 |
| Resolution | FIXED |
| Resolved on | Oct 09, 2006 18:26 |
| Version | 1.00beta |
| OS | Linux |
| Architecture | PC |
| Attachments | [rlctest.sac](/uploads/d5fa27ca911068c681fbaa109b35fb60/rlctest.sac), [RandLC.sac](/uploads/31388a23a10c859de8b59cb81e9abc2e/RandLC.sac) |
## Extended Description
<pre>Function rlctest breaks the compiler:
use RandLC: all;
int rlctest( int i, RandLC &rlc) {
j = _toi_S_( randlc(rlc));
return(j);
}
int main() {
tran = 314159265d; amult = 1220703125d;
rlc = create_randlc( tran, amult);
i = rlctest(2, rlc);
delete_randlc( rlc);
return(i);
}
with error message "undeclared identifier"
** 18: Invoking C compiler: ...
gcc: unrecognized option `-R.'
a.out.c: In function `SACf__MAIN__rlctest__i__SACt_RandLC__RandLC':
a.out.c:360: warning: assignment makes integer from pointer without a cast
a.out.c:360: `SAC_arg_1__desc__p' undeclared (first use in this function)
The error message does not appear if we put the reference argument to the front
of the argument list, or remove it entirely.</pre>BugZillaBugZillahttps://gitlab.sac-home.org/sac-group/sac2c/-/issues/1016global objects inhibit fold functions2021-05-26T20:27:01ZSven-Bodo Scholzglobal objects inhibit fold functions| | |
| --- | --- |
| Bugzilla Link | [306](http://bugs.sac-home.org/show_bug.cgi?id=306) |
| Created on | Oct 12, 2006 10:59 |
| Version | 1.00beta |
| OS | SunOS |
| Architecture | PC |
## Extended Description
<pre>This bug is a co...| | |
| --- | --- |
| Bugzilla Link | [306](http://bugs.sac-home.org/show_bug.cgi?id=306) |
| Created on | Oct 12, 2006 10:59 |
| Version | 1.00beta |
| OS | SunOS |
| Architecture | PC |
## Extended Description
<pre>This bug is a conceptual problem but feels very much wrong...
if we introduce a print in a function used within a fold, the
subsequent change in signature inhibits the use as fold and thus leads
to a strange error message: ...function not found.....
Since all overloadings of a function are treated uniformly, the
offending print may even occur in a fold-irrelevant overloading only.
I am not sure how to handle this problem nicely:-(
I would assume that, in principle, this problem could be solved
in a rather complex modification of fold-wls using propagate or so.....</pre>BugZillaBugZilla