Project

General

Profile

Bug #850

Some types of selections don't work correctly when compiled in double precision

Added by Teemu Murtola over 5 years ago. Updated about 4 years ago.

Status:
Closed
Priority:
Normal
Assignee:
Teemu Murtola
Category:
selections
Target version:
4.5.6
Affected version - extra info:
4.5-4.5.5
Affected version:
4.5.5
Difficulty:
uncategorized
Close

Description

There is a bug in handling integer to floating-point parameters for selection methods, which causes some types of selections to behave erratically if Gromacs is compiled in double precision. Or more accurately, if sizeof(real) != sizeof(int).

This only affects g_select, and possible user code written using the new trajectory analysis framework.

Associated revisions

Revision f35e2ca8 (diff)
Added by Teemu Murtola over 5 years ago

Fix selection parsing in double precision.

Fixes #850.

Change-Id: Ifd0bf1f5401ca4d62571c8c2fc39e67cac537ecf

History

#1 Updated by Teemu Murtola over 5 years ago

  • Status changed from New to Closed

Fix has been reviewed in Gerrit and merged into release-4-5-patches.

#2 Updated by Teemu Murtola about 4 years ago

  • Category changed from analysis tools to selections
  • Affected version set to 4.5.5

Also available in: Atom PDF