Base64 Encoding in XMI Export

[SvenK]

Hello,

the baseline feature is basically an XMI export of the current model. We use the XMIs to compare different baselines using external tools. However, we faced the problem that Base64 encoded data (e.g. Linked Document) always changes slightly, although no changes where done. It is always the first and the two last lines of the encoded data which changed. Is there some kind of timestamp stored? How can I avoid this?

How is the data stored? I thought it would be plain RTF. However, I can't open a file containing the decoded data.

Another strange behavior I got with the xmi.id of stereotypes. The ID of e.g. stereotype "master document" always changes also when creating a baseline.

Any hints on this?

Sven

[qwerty]

I think this is connected with zipping the XMI contents. I once asked a similar question (but can't remember the context) where I was told that from a Sparxian. What you have to do is to unzip the base64 and compare the raw file.

Just if you haven't done that: either unzip via some self-made program or via Notepad++: copy/paste the base64 content into an empty buffer; select all; choose decode base64 from the mime tools; save as tmp.zip; open the zip and look into the contained file str.dat.

q.

[SvenK]

Thanks. That helped. The difference is due to different modify times of the str.dat file archived in the zip.

Now I have to figure out how to avoid changing GUIDs for (default) stereotypes.

[qwerty]

The difference is due to different modify times of the str.dat file archived in the zip.

That's it! Maybe Sparx could fix that to a certain date (say the first release date of EA or Geoffrey's birthday). It isn't used anyway.

q.