Hi David,

> Have you had any corruption issues with this technique?

Not as a result of an automated merge, if that was the question. However, I always view differences before I check in to avoid any bad surprise later one.

I did have corruptions when I started working on the tool, because of invalid characters in the XML file. I fixed them except for reports and labels. That's why currently GenCode skips those.

Paul McNett also has a two way tool. His approach is slightly different in that he generates text files right after you save the file and before you edit it: http://paulmcnett.com/scX.php.

> How well does the merging work?

Fine for me. However, you should check in frequently. If CVS cannot resolve a conflict, it inserts the the repository version and your code into the file with a comment. Huge blocks of conflicts are hard to deal with,. because you have to find out why you have a conflict in the first place and which one is the correct code.

> How much overhead is this in terms of your time?

GenXML is pretty fast on my machines. It takes a few seconds to convert all file. GenCode is a bit slower, maybe three times the time of rebuilding a project. So far, nobody found this too far overhead, it's more the check in comments that I have to fight for. <g>