MD5Sum should ignore whitespace!

One of my biggest headaches right now is if I reformat the liquibase xml file, then when I try to run the maven build next it dies because the MD5Sum doesn’t match the one in the database. I really think that liquibase should strip whitespace when calculating MD5Sums so that you can reformat the xml file without having builds start failing.

For the most part it does.  The only place where it preserves whitespace is text within a tag.  We do that because depending on the text in there, the whitespace may matter.  That said, I will make sure that for 2.0 (where we change how we compute md5sums and are breaking some backwards compatibility) we tell the XML parser to strip any insignificant whitespace according to the XML spec.


Is this still pending? We’re currently experiencing exactly this same problem.