On Thu, 2008-02-21 at 20:01 +0200, Tsantilas Christos wrote: > I believe, I solved (most of?) the problems in astyle. > I am attaching the new version of the formatter.pl > I test it only using the astyle version 1.21.
Christos, Thanks a lot for polishing this! Any objections to committing Christos' astyle wrapper and then applying it to HEAD and selected branches? If there are no objections, let's set a date for this. How about March, after 3.0.2 is released, with a final 7-day warning? Or is it better to reformat before releasing 3.0.2 so that it is easier to backport future fixes? > Also I am attaching a simple script I used to check the formated code. > This scripts removes any space, tabs and newlines from the original file > and the formated file, and compare their md5 signatures.This script > requires the tr and md5sum utilities. I test it only on a linux system. > > This script can not detect bad formated code but maybe can detect bad > modifications (eg source code blocks removal) on source code. > > Running this script I found 5 files with different md5 signatures but > all of these files had formating modifications similar to the following: > From > if (i<t) /*A comment*/ > { > ... > } > To > if (i<t) { /*A comment*/ > ... > } Ha! Who would have thought that such transpositions are possible? I guess we need to strip not just whitespace, but comments as well (which is more difficult) for the MD5 test to work 100%. I would not spend time on that now. Thank you, Alex.