errors while using redistance.h file

[Kaustubh Thakurdesai]

Hi all

Recently I have been trying to rectify some cases of level set method in Basilisk. As i understand, redistance.h is the maintained version, for performing the re-distancing of the level set, while LS_reinit.h is the older version, which is no longer maintained.

I am encountering a peculiar issue here. When I simulate the cases in my laptop, using redistance.h, they run perfectly fine, while when  i try to run them on the Basilisk website, the redistance.h gives a bunch of errors, while LS_reinit.h runs without any errors on the website.

Has anyone encountered a similar issue? Any information regarding this would be helpful.

Is there something trivial that I am missing here?

Thanks

Kaustubh

[Stephane Popinet]

> I am encountering a peculiar issue here. When I simulate the cases in my
> laptop, using redistance.h, they run perfectly fine, while when  i try
> to run them on the Basilisk website, the redistance.h gives a bunch of
> errors, while LS_reinit.h runs without any errors on the website.

Please send the links to the running (with LS_reinit.h) and crashing
(with redistance.h) examples.

Stephane

[Kaustubh Thakurdesai]

Dear Stephane

My apologies for the delay in reply. The reason being I noticed that since past couple of days the redistance.h is running properly on the Basilisk website as well now while it was giving errors when I posted the message (An example is one of the cases from Alexandre Limare's sandbox (https://basilisk.fr/_edit/sandbox/alimare/1_test_cases/reinit_1D.c) or  (https://basilisk.fr/sandbox/alimare/1_test_cases/reinit_circle.c).

However, I am still facing the broader issue of a code running properly on my workstation and having errors on basilisk website. One example is the meanflow_3D case again from Limare's sandbox (https://basilisk.fr/sandbox/alimare/1_test_cases/meanflow3D.c).I get exactly the same reference result shown in the sandbox page, when running on my workstation. 

But when i simulate this on Basilisk website, I get the following error:

/home/basilisk-src/basilisk/src/grid/tree-common.h: In function ‘main’:
src/grid/tree-common.h:160:1: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
DT2 0.00012207
prerefine nbcells 10648
prerefine nbcells 38536
prerefine nbcells 39992
prerefine nbcells 40160
nbcells40160
iteration 0
src/utils.h:178:error: [Thread debugging using libthread_db enabled]
src/utils.h:178:error: Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
src/utils.h:178:error: [Detaching after vfork from child process 2454178]
src/utils.h:178:error: Program received signal SIGFPE, Arithmetic exception.

On the other hand , I get the following result attached without any errors or warnings from my workstation.

I would be grateful for any possible information about this issue with code simulating on my workstation but having errors on the Basilisk website.

Thanks & Regards

Kaustubh

[Stephane Popinet]

Hi Kaustubh,

> My apologies for the delay in reply. The reason being I noticed that
> since past couple of days the redistance.h is running properly on the
> Basilisk website

Yes, the problem was that an old redistance.h file was left in the
directory and hadn't been cleaned up somehow. So that was interfering
with the default src/redistance.h. Removing the file fixed this problem.

> However, I am still facing the broader issue of a code running properly
> on my workstation and having errors on basilisk website. One example is
> the meanflow_3D case again from Limare's sandbox (https://basilisk.fr/
> sandbox/alimare/1_test_cases/meanflow3D.c).I get exactly the same
> reference result shown in the sandbox page, when running on my workstation.

OK, I just fixed the problem which was that the (rather disgusting)
formatting of the Makefile here:

https://basilisk.dalembert.upmc.fr/sandbox/alimare/1_test_cases/Makefile

was confusing the running script (only in "edit" mode on the website),
which as a consequence was just ignoring the option:

meanflow3D.tst: CFLAGS += -fopenmp -DTRACE=2

So now this case also runs on the server. Note however that the results
look completely wrong... (unlike the movie you sent which looks OK). If
I run it on my local machine, I get the same (wrong) result as on the
server (so different from your movie), so I think you are not running
the same versions of Basilisk and/or other codes on your local machine.

Are you using darcs to get the full wiki? Did you make sure to update
your basilisk version using darcs?

Note also that if you comment out the line above in the Makefile on your
local machine, you will also be able to reproduce the crash you had
before when running on the website. There is no reason that the code
should crash without these options. Given these two things (plus the
rather scary warning you have about pointer sizes) I suspect that the
code is quite buggy...

cheers,

Stephane