Cameron Hart at Flow Software has just blogged about a change he identified in MIDAS.DLL to significantly improve the performance (and capacity) of client data sets when handling large volumes of records and BLOB fields. Basically he seems to have taken the work previously done by Andreas Hausladen and applied it to BLOB’s.
The performance improvements were hugely significant, but Cameron is quite open about the fact that he’s primarily a Delphi developer, and so is interested in any comments or observations from those perhaps more familiar with C/C++ and especially anyone more intimately familiar with the inner workings of the MIDAS.DLL that might be able to spot any issues in his changes that perhaps aren’t apparent to him.
He’s published the changes he made on the blog post. The formatting of the code isn’t great thanks to using a hosted installation of WordPress which presumably doesn’t allow syntax highlighting plugins to be installed, so you might want to copy/paste his code into a more C/C++ friendly editor for detailed perusal.
In the interests of full disclosure I should mention that Flow Software is my “Daily Planet”.