Archive from user: Chris Neale I am not sure if this is an error in gmx distance or simply an indication that the program doesn’t fail gracefully when I do something idiotic. I have attached a .pdb file and a script for running gmx distance commands that works on some selections but gives a memory allocation error on other selections. I believe that all selections should be valid. More information is available here: https://mailman-1.sys.kth.se/pipermail/gromacs.org\_gmx-users/2016-November/109623.html *(from redmine: issue id 2086, created on 2016-12-02 by gmxdefault, closed on 2017-01-20)* * Changesets: * Revision c848f4a4b18eed9cd36d97a59d5d97f0cdd5ad4e by Teemu Murtola on 2016-12-07T01:03:01Z: ``` Fix use of position variables with plus/merge If a selection contained a position variable (e.g., 'com of ...') that was used more than once, and at least one of those uses was with plus/merge, there were out-of-bounds memory writes. This was caused by the internal position structure not getting fully initialized. Incomplete initialization happens in all contexts with such variables, but only plus/merge (and possibly permute) actually use the values that remained uninitialized, which caused them to incorrectly compute the amount of memory required to store the result. Fixes part of #2086. Change-Id: I016e796db268a11d557309935c02cbd1bc79a83c ``` * Revision b7817e2da7e2e4df1292e7bc3f4f5414a1a77999 by Teemu Murtola on 2016-12-08T16:21:56Z: ``` Fix possible memory error with long selections If a selection was more than 1000 characters long and there was a whitespace exactly at the 1000 point, a buffer overflow could occur. Replaced the buffer with std::string, simplifying the code significantly. Update the generated code to use a newer flex, which also removes the need for some suppressions. Should fix #2086. Change-Id: I56513bcf5ee99f05ce144461740d0f868be10186 ``` * Uploads: * [gmxDistanceErrorFiles.tgz](/uploads/4aad370a13054c97685256693d92827b/gmxDistanceErrorFiles.tgz)