5

Closed

SSIS: smart source control diff

description

As the underlying XML files are structured and formatted currently, you get a bunch of messiness when you do a diff of two versions of an object (whether it be a cube, dimension, SSIS package, etc.) I think it might be worthwhile to code a "smart diff" which gets rid of the clutter and makes it easier to quickly see what's changed between versions.
 
Here are a couple of things off the top of the head that could be improved...
 
  • In many objects, all attributes of an XML tag are listed on one row, making it difficult to tell which changed. We could split the attributes out onto their own row and indent them correctly to make diffs more readable.
     
  • In an SSIS package, the underlying XML is a complete mess in terms of trying to do a diff on versions. We could restructure it to make it more readable and meaningful for a diff.
Closed Nov 16, 2007 at 10:27 PM by dgosbell

comments

dgosbell wrote Jan 17, 2007 at 5:00 AM

MSDN had an xmlDiff sample that might be useful - it does a symantic diff. ignoring things that are not significant like whitespace and node ordering, but I agree that there is still a need to do some pre-processing to clean out other things like DSV object positioning which does not have any effect on the final result.

dgosbell wrote Jan 17, 2007 at 5:01 AM

sorry forgot to add the link to the sample http://msdn2.microsoft.com/en-us/library/aa302295.aspx

dweinst wrote Sep 11, 2007 at 1:27 PM

Might be worth considering going down the AMO route rather than relying on XML diff for the SSAS side...