00096: Dirty diff result using phpdiff.php of PmWiki-1.0.11
Description: This phpdiff.php function make empty lines to be compared either, thus may give out dirty diff result. Following is an example:
Old text is:
title ---- subtitle1 text1 ---- subtitle2 text2
And new text is:
title ---- new subtitle new text ---- subtitle1 text1 ---- subtitle2 text2
Then the phpdiff function give diff result like:
Changed line 5 from: subtitle1 to: new subtitle Changed line 7 from: text1 to: new text Changed line 11 from: subtitle2 to: subtitle1 Changed lines 13-19 from: text2 to: text1 -------------------------------------------------------------------------------- subtitle2 text2
My patch for this is just simply make phpdiff function not to compare empty lines.
I add such lines after the line 50 of the phpdiff.php file:
if ($t1[$s1]=='' || $t2[$s2]=='')
{ $s1++; $s2++; continue; }
Agreed, except that the above patch is a little too aggressive about avoiding matches--i.e., if $t2[$s2] is blank then it doesn't even consider if $t1[$s1] might be a worthwhile (non-blank) match and the shortest distance.
I've implemented the above scheme differently--by simply not including blank lines as valid targets for the shortest-distance search (i.e., don't include them in the reverse-index array in the first place).
Thanks for the excellent analysis and suggestion! --Pm