[Psrfits] CFITSIO bug

[George.Hobbs at csiro.au]

Thanks heaps Willem for finding this.  Do you have any understanding why we've been using PSRFITS for ages with e.g., the DFB4 without (seemingly) any issues at all?

Regards,
 George

________________________________________
From: Psrfits [psrfits-bounces at lists.pulsarastronomy.net] on behalf of Willem van Straten [vanstraten.willem at gmail.com]
Sent: Sunday, 4 March 2018 7:23 PM
To: psrchive-users at lists.sourceforge.net; psrchive-developers at lists.sourceforge.net; psrfits at lists.pulsarastronomy.net
Cc: everyone at lists.pulsarastronomy.net; PPTA
Subject: Re: [Psrfits] CFITSIO bug

Hi everyone,

The latest version 3.430 of the CFITSIO library includes a bug fix that addresses the ffmvec problem encountered last year.  It is available at

https://heasarc.gsfc.nasa.gov/fitsio/fitsio.html

Cheers,
Willem


On 28 September 2017 at 21:34, Willem van Straten <vanstraten.willem at gmail.com<mailto:vanstraten.willem at gmail.com>> wrote:
Hi all,

There is a fairly major bug in the CFITSIO library that can lead to memory violation.

In most cases, PSRCHIVE programs that use the library will complete without any sign of error; however, in some cases, the program will either fail with a misleading error message or enter an infinite loop during which the output PSRFITS file grows until the process is killed or the file system is filled.

I'm surprised that the bug does not appear to result in data corruption more frequently, and that it was not noticed earlier.

I've sent a patch to the HEASARC CFITSIO Help Desk.  This patch is also distributed with the latest version of PSRCHIVE and there are some notes on how to install it at

http://psrchive.sourceforge.net/trouble/ffmvec

Cheers,
Willem