-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
R shuts down when trying to run dwnominate on mac #1
Comments
I'm pretty sure this is an obscure compiler optimization issue, caused by macs using an older version of the gfortran compiler. I think it can be fixed by using a lower optimization setting. To do that, first download this repository. Then create a file called
Now open the terminal and run R CMD INSTALL path/to/dwnominate-master If you can confirm that it works after that I'll try to set up a Makevars here to make that happen automatically. Should've done that a long time ago. Oh! And can you please tell me what you get when you run |
Thank you for the fast answer! I tried this, but I get the following errors in the terminal when running the INSTALL command:
I am not sure, whether I did the part with the Makevars-file right. Is this supposed to be a .f Fortran file or does it have to have some other extension than The
Thank you very much for your help! |
Just name the file You actually have a newer gfortran version than I do. Strange. |
aah, thank you. I have gotten the following error in the terminal now, though:
Does the I am sorry, as mentioned above, I am new to Fortran and its surroundings so my questions might be somewhat arbitrary sometimes... |
I got the same error earlier when the last line of I don't know all of the details about how I'm not so big on fortran myself-- I wrote this R package so I could avoid it. :) |
Alright, now it installed something, but after trying to run it in R it still aborts after the W-NOMINATE estimation. By the way, it creates all the files it needs to create (e.g. "DW-NOMSTART.DAT" and "DWNOM21.DAT") and aborts after that.
|
Can you send me the data and script you're using? Then I could see if it runs for me, suggesting a compiler issue. If I get the segfault too then the problem is probably somewhere else. In the meantime you can try replacing |
I have sent it to your email. I tried the --O0 version as well, which did not work as well... |
Alright. The example worked for me. (You were testing this with just the two sessions, right?) I'm using gfortran 5, so if you're able to switch your gfortran version that's worth a try. I'll experiment with gfortran 6 to see if I can find the issue, but I have no idea how long that'll take me. |
Alright, good to hear that it at least could work with my data. ;) Yes, I was testing it with two sessions, I have data for the last 20 years. |
Which macos do you use by the way (I use 10.12.5)? Maybe it has something to do with that? Or with the R version (I use 3.4.1 now). |
I'm on Linux (Ubuntu 16.04), and I also have R 3.4.1. My specific gfortran version is 5.4.0. |
Oh, so if you use Linux, it might be a problem with mac? I think gfortran 5.4 does not exist for mac, at least not here. |
OK, I got some advice from my local fortran guru. Add You can also try running |
Alright, this is what the terminal printed when using
I'm not sure whether it is that much different from the output I got before. |
This is interesting! I was not able to change the ulimit (it might not be possible on mac), but I nevertheless ran everything in an R session in the Terminal. And this did not just break down like R does, but it gave me a specific error message:
Maybe this helps you find out where the problem is? |
Can you send me your For example, here's what my
|
I sent it to you by email. I also attached all the other data that |
I didn't get the email, can you send it again? |
I have it on good authority that the code will run correctly if you get the version of gfortran described in this stackoverflow question. I'm going to wait to hear more from the gfortran people before deciding what to do with this. And I'll try to get a Mac for testing so we don't have to go back and forth via email anymore. |
yeah! it did work with this gfortran version! thank you very much!
… Am 22.07.2017 um 00:35 schrieb Will May ***@***.***>:
I have it on good authority that the code will run correctly if you get the version of gfortran described in this stackoverflow question <https://stackoverflow.com/questions/23916219/os-x-package-installation-depends-on-gfortran-4-8>.
I'm going to wait to hear more from the gfortran people <https://gcc.gnu.org/ml/fortran/2017-07/msg00042.html> before deciding what to do with this. And I'll try to get a Mac for testing so we don't have to go back and forth via email anymore.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub <#1 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AcqUUDWaLeksy0Ze26fKGHEZsoxhk6M2ks5sQSeagaJpZM4OTOdt>.
|
Hi,
I am trying to run the dwnominate function in Rstudio (Version 1.0.143), running R version 3.4.0 on a mac (macOS 10.12.5) machine. I use rollcall data of Swiss MPs. It first successfully computes the W-NOMINATE scores, but then R always shuts down (I attached a screenshot of the error message):
Is there a problem with the fact that I try to run this on mac? Do I have to first Install the Fortran program of Poole on my machine or does this get done by default when I install the R-package? And if I have to first install the Fortran program, is this even possible on a mac machine?
I am very new to working with Fortran, I hope, someone can help me.
Thank you!
David
The text was updated successfully, but these errors were encountered: