Tuesday, January 30, 2007 - Tue, 30 Jan 2007

After sending out my e-mail to let people know about the new version, I received two interesting e-mails back about the same thing from two different people. What they both had to say really surprised me.

Both have very large GEDCOMs, upwards of 100,000 names. Behold still does not have an easy time with files of this magnitude. My idea was that if Behold takes more than 10 or 15 seconds to load a file and produce its Everything Report, then it’s too long for most people to wait. But they told me otherwise.

In fact, they’d both be willing to wait for it to load, as much as 6 hours if they had to - as long as all the viewing and moving around in the Report would then be fast - which in Behold it is.

What I didn’t realize is how long other programs take to read in large GEDCOMs. One of the people told me that for their very large dataset, RootsMagic imports it in a few minutes, Legacy takes half an hour and TMG took hours and finally crashed. I didn’t realize other programs took so long, simply to import the data.

With regards to Behold, I was told it took hours, gobbled up more than half a gigabyte of memory and seemed to hang. Behold was “red-lining” (as its memory dials were indicating no memory left) and when that happens, page swapping of memory takes over and you can just about forget it.

But the difference here was that the other programs were just loading the GEDCOM data. Behold easily loads the GEDCOM data of very large GEDCOMs. It’s the generation of a 30,000 page Everything Report that bogs Behold down.

So for a true comparison, I downloaded a trial version of RootsMagic and tried it on a 21 megabyte GEDCOM containing 34,000 people that I have for testing. It took RootsMagic two and a half minutes to import the GEDCOM file. By comparison, Behold took only 13 seconds and required 58 MB of memory to load the GEDCOM. Yay for Behold, as its GEDCOM loading must be okay. Then Behold continued and took another 140 seconds and required 285 MB more memory to generate the Everything Report. What I haven’t done yet is work to optimize the building of the Report, and as I’ve recently mentioned in this blog, there is lots of potential improvement.

But the e-mail comments I received made a point as I now realize that Behold may in fact be a tool that people with very large datasets could want to use, even if they had to wait a bit. That’s good to know!