crash
so i have just crashed openscad for the first time.
i think its a circular dependency issue
in file attached "Connectors.scad" i accidentally left a number 1 where
i had copy passed some data and openscad just crashed with no
explanation.
re opening the file just instantly crashes openscad again
so i opened the file in gedit and found the problem witch i have
corrected in the attached file "Connectors (copy).scad" and im now back
in action.
but i wondered is anyone was interested in looking why it actually
crashes openscad rather than producing a warning or error
the offending line is line 168 and 177 of Connectors.scad
changing the lines from
function SocketF1_Depth(depth) = depth
+(Socket_Thickness1()*2)+(Socket_Thickness2()*2)+(Socket_Clearance()*4);
function SocketF1_Depth() = SocketF1_Depth(Socket_Depth1());
to
function SocketF_Depth(depth) = depth
+(Socket_Thickness1()*2)+(Socket_Thickness2()*2)+(Socket_Clearance()*4);
function SocketF1_Depth() = SocketF_Depth(Socket_Depth1());
fixes the problem
the rest of the attached files are dependences in case anyone tries to
re produce the crash
Hint: if you want developers - of any program - to try to diagnose
something, it's best to do whatever you can to provide a test case that
they can readily run, and reduce it to the smallest case that
demonstrates your problem. Here you have given us programs that we
can't immediately run (because they include references to /home/jay/...,
and one of the files that appears to be required (Fuse Holder.scad) is
missing, and the total of the three files is somewhere around a thousand
lines.
Exactly what version of OpenSCAD are you running, on what platform?
When you say "crashed", do you mean that the window disappeared, or that
it stopped responding? If on a UNIX derivative, was there a core dump?
If I have "stop on first warning" turned off, I get
WARNING: Too many unnamed arguments supplied in file
Connectors.scad, line 173
five times, and then OpenSCAD becomes unresponsive.
The problem is indeed with line 173:
function SocketF1_Depth() = SocketF1_Depth(Socket_Depth1());
which is infinitely recursive - it calls itself, with no way to stop -
and has the call have one argument while the function has none.
Normally OpenSCAD detects infinite recursion, but I suspect that here
the checks may be optimized out of existence by tail recursion
optimization.[*]
[*] I've never bothered to fully understand the cases that qualify
as tail recursion, so I could easily be wrong here.
If I'm right, the result is more or less straightforwardly an infinite
loop and, since there's no way to stop a running OpenSCAD program, it's
fatal.
Probably the most relevant open bug report is #1195
https://github.com/openscad/openscad/issues/1195, that previews should
run in a separate thread so they do not hang the UI.
Workaround: other than fixing the infinite recursion in your program, I
suggest that you turn on "Stop on first warning" (in
Edit/Preferences/Advanced). I suspect that you also have Design /
Automatic Reload and Preview turned on; unless you use an external
editor I would suggest turning that off.
On 12/22/2022 3:56 PM, Jordan Brown wrote:
Workaround: other than fixing the infinite recursion in your program,
I suggest that you turn on "Stop on first warning" (in
Edit/Preferences/Advanced). I suspect that you also have Design /
Automatic Reload and Preview turned on; unless you use an external
editor I would suggest turning that off.
It occurs to me that "Stop on first warning" is a workaround only in
this particular case, where there's an error. If there was undetected
infinite recursion without an error, that wouldn't help.
Turning off "Automatic Reload and Preview" would help - and you could do
it in an empty OpenSCAD, without loading the problematic file - so that
you can load a file that causes problems without triggering those problems.
Note also that OpenSCAD saves a backup copy before every run; look at
the console log for the name of the backup file.
Can't you just open a new file, turn off Auto reload and then close
openscad and open the problem file?
On Fri, 23 Dec 2022 at 01:24, Jordan Brown openscad@jordan.maileater.net
wrote:
On 12/22/2022 3:56 PM, Jordan Brown wrote:
Workaround: other than fixing the infinite recursion in your program, I
suggest that you turn on "Stop on first warning" (in
Edit/Preferences/Advanced). I suspect that you also have Design /
Automatic Reload and Preview turned on; unless you use an external editor I
would suggest turning that off.
It occurs to me that "Stop on first warning" is a workaround only in this
particular case, where there's an error. If there was undetected infinite
recursion without an error, that wouldn't help.
Turning off "Automatic Reload and Preview" would help - and you could do
it in an empty OpenSCAD, without loading the problematic file - so that you
can load a file that causes problems without triggering those problems.
Note also that OpenSCAD saves a backup copy before every run; look at the
console log for the name of the backup file.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
On 12/23/2022 1:26 AM, nop head wrote:
Can't you just open a new file, turn off Auto reload and then close
openscad and open the problem file?
Yes. You could do that once, and not have to do it again or, if you
like auto-refresh, you could do it and undo it each time that you have
such a problem.
I don't really see much of a win for auto-refresh if you use the
built-in editor, so I keep it off.
For what it's worth, OpenSCAD does catch infinite recursion even if tail
recursion is involved; it's just that the limit is much higher.
function recurse(i) =
let (dummy= (0 == i % 100) ? echo(i) : 0)
recurse(i+1);
echo(recurse(0));
This errors out after 1,000,000 recursions.
If I change the last line to:
0+recurse(i+1);
it errors out after 7,200 recursions.
The condition for tail recursion is that the recursion call has to invoke
the function while doing nothing else. The "0+" is something else, so it
disqualifies the function for tail recursion.
On Fri, Dec 23, 2022 at 12:27 PM Jordan Brown openscad@jordan.maileater.net
wrote:
On 12/23/2022 1:26 AM, nop head wrote:
Can't you just open a new file, turn off Auto reload and then close
openscad and open the problem file?
Yes. You could do that once, and not have to do it again or, if you like
auto-refresh, you could do it and undo it each time that you have such a
problem.
I don't really see much of a win for auto-refresh if you use the built-in
editor, so I keep it off.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
On 12/23/2022 12:20 PM, Father Horton wrote:
For what it's worth, OpenSCAD does catch infinite recursion even if
tail recursion is involved; it's just that the limit is much higher.
Interesting. I spoke too fast about this being straightforward infinite
recursion. The hang here is somehow tied up in the fact that there's
also an argument error.
This program crashes:
function foo() = foo(0);
echo(foo());
But these two programs do not; they get a "recursion detected" error:
function foo(x) = foo(0);
echo(foo());
function foo() = foo();
echo(foo());
I've filed
Infinite recursion with too many function arguments hangs #4459
Well spotted. I dident realy look at why i just picked through the file untill i worked out how to make it run again. Sorry againSent via the Samsung Galaxy S7, an AT&T 4G LTE smartphone
-------- Original message --------From: Jordan Brown openscad@jordan.maileater.net Date: 12/23/22 10:15 PM (GMT+00:00) To: OpenSCAD general discussion Mailing-list discuss@lists.openscad.org, Father Horton fatherhorton@gmail.com Subject: [OpenSCAD] Re: crash
On 12/23/2022 12:20 PM, Father Horton
wrote:
For what it's
worth, OpenSCAD does catch infinite recursion even if tail
recursion is involved; it's just that the limit is much
higher.
Interesting. I spoke too fast about this being straightforward
infinite recursion. The hang here is somehow tied up in the fact
that there's *also* an argument error.
This program crashes:
function foo() = foo(0);
echo(foo());
But these two programs do not; they get a "recursion detected"
error:
function foo(x) = foo(0);
echo(foo());
function foo() = foo();
echo(foo());
I've filed
Infinite recursion with too many function arguments hangs #4459
On 12/23/2022 2:14 PM, Jordan Brown wrote:
Interesting. I spoke too fast about this being straightforward
infinite recursion. The hang here is somehow tied up in the fact that
there's also an argument error.
Hans Loeblich writes in the Github issue:
This is entirely due to the "Error Log", which seems to take linear
time to append each error. This is why we should probably have some
configurable "Max Error Limit" in preferences (with a default of
maybe 100).
This could also potentially replace the "stop on first warning"
option (by setting the value to 1).