Big Mean File Processing

What a couple of weeks it has been since the launch of Leopard..

I’m not sure whether I’m the only one to find this, but it does look like one of the buggiest Apple releases in recent history..

By in large, a new operating system release should have zero impact on existing programs. This time over, however, it looks like lots and lots of programs have been broken all over the place..

Of course, as a developer I spend most of my time in Apple’s development tools, namely XCode and the brand-new “Instruments”. I’m sad to say it, but the current XCode 3.0 is a no more than an alpha release. The editor hangs, the snapshot feature doesn’t work (force quit), half the time after running a program in debug mode, you need to quit XCode to “reset” it.. it will be great when it’s finished, but it has been released at least a year too early 🙁

Time Machine is great, but lacks configurability and feedback. Every hour it will pretty much freeze the machine and backup 360Gb of data (build directories do not seem to be excluded from backups), the only indication that you have that it’s busy is that the external hard disk starts making a racket, there’s no easy way of delaying a backup, etc, etc.

My own applications have also been hit by Leopard problems despite having tested them on various beta builds of the new operating system.

For A Better Finder Rename, the migration problems were largely due to changes (and apparently new bugs) in the Finder and in AppleScript. For MacBreakZ there are some oddities to do with Spaces (which does behave in some fairly strange ways).

At the moment, however, it seems that it’s the Big Mean Folder Machine that has been the hardest hit. This is largely due to some very nice progress on Mac OS X’s built-in Core Data database engine. It’s much faster, but it also seems to leak memory fairly badly where it didn’t before. As a consequence, “The Big Mean Folder Machine” appears to be both a lot faster (great!) and lot memory hungrier (no!) on Leopard than it was on Tiger..

Unfortunately the vastly increased memory usage on Leopard makes the whole application crash when it runs out of memory. Luckily this doesn’t happen during your garden variety 6000 file re-organization, but can quickly become a problem when you’re dealing with tens or hundreds of thousands of files. Well using Core Data was supposed to provide “The Big Mean Folder Machine” with fantastic scalability rather than make memory a problem!

I have therefore spent much of the last week optimizing “The Big Mean Folder Machine” for Leopard by manually testing each bit of code for its memory usage.. this is very tedious work, but at least it seems to be working:

My current test version now deals gracefully with half a million files and has a very small memory foot print even when it’s working flat out..

I hope to have a new version of “The Big Mean Folder Machine” that will gracefully deal with over a million files out very soon, so expect “The Big Mean Folder Machine” 1.3 very soon.

This is also good news for the upcoming version 8 of A Better Finder Rename, which will inherit much of The Big Mean Folder Machine’s processing pipeline.

If anybody is stuck with the current version of “The Big Mean Folder Machine”, please just email me and I can send you a development copy via email.

Best regards,

Frank

2 thoughts on “Big Mean File Processing

  1. Yeah. Although a front end Web person, not a developer, I too find Leopard, 10.5.0 and 10.5.1 ‘not nice’ compared to Tiger in terms of it’s good behaviour.

    Had to post a comment also just to say that I think MacBreakZ is excellent, have used it for a couple of months – every busy Mac user ought to own a copy. And A Better Finder Rename is just perfection-for-file-renaming. Thank you lots Frank for these truly valuable tools. Cheers, -Alan

Comments are closed.