Thanks, Bruce. I'm moving this thread back to the ifeffit board:

On Thursday 31 March 2005 18:41, you wrote:

...

Incidentally, trying to shrink down the file has made me notice there's a lot of leftover stuff in the zip file--it looks like every data file I've ever used for chi data is still in there, as well as an old fit for each of those. Artemis almost never crashes any more (yay!), so those aren't a legacy of crashes--it just appears that every time I change data files it leaves the old one (and the associated fit) sitting in the project file. Since I sometimes use the same model with lots of different data, that starts to add up.

Before I actually split Artemis into two branches, I was already starting to think about the concept of maintaining a history of all fits done. The presence of all that stuff makes no sense in the context of the stable branch, but it is much more sensible in the development branch.

The data should all be chi(k) data -- not so big and it compresses well. Project files are big because the feff calc, not because of the data.

The example I sent you had about 300 k of the zipped project file in old chi data, and the total file is 3 M, so yes, most of it in this case is FEFF calculations. I'm concerned, though, that the former number can only go up. Here's what I think is a reasonable hypothetical: someone has a really stable Artemis model that they like for their materials, and they use it over and over again on various samples. Unlike the file I sent you, it might also use a fairly small number of FEFF paths. Combine that with the fact that the time used by Autosaves is of course quite dependent on the size of the project file (and maybe the number of files in it? I'm not sure how the zipping algorithm works in that respect), and it becomes potentially annoying. I don't consider this a high priority issue, but it seems like something worth thinking about.

Another future reason for hanging onto all the data is batch functions. Suppose you had a temperature series and wanted to fit each of the scans with the same model. Wouldn't it be cool to tell artemis-of-tomorrow to fit all the data and then show the temperature dependence of one of the sigma^2s?

Yes, it would be very cool. I guess I'm just asking for a "clear history" option or something like that. In the mean time, I'm happy to do this manually if I know it won't confuse Artemis. If I just root around in the zipped project file and remove the old .chi and .fit files, will that confuse Artemis in any way? --Scott Calvin Sarah Lawrence College