Our customers have spoken and we listened. We received a bunch of requests to speed up and optimize the memory usage of our trends feature in NCover 3. For more information on trends please click here.
NCover v3.1.4 includes a brand new binary data format for coverage trends. What does that mean for you? It means that the entire set of trend data no longer needs to be loaded into memory any time you work with trends. The time and memory required to append a new coverage file to the trend file has been reduced by as much as 80% and the file size by 50-66%. It also means that we are able to read any piece of trend data from any location in your coverage file in a fraction of the time it took with the existing format. Associating a trend file to a coverage file is now an instantaneous operation. No longer will you watch the “Loading Trends” dialog.
With changes come new behaviors. By default, if the coverage file path does not exist, we will create one for you and it will be empty until trends are added to the file. We will also require you to use the new binary format for any new trend files. You will no longer be able to create a new xml based trend file from scratch. Your existing xml based trend files will continue to work, but we highly recommend converting them to the new format using the new converstion tool that we will talk about next.
NCover v3.1.4 ships with a new tool called ncover.upconvert. The experience of moving to NCover 3 taught us that people do not want file formats changing by magic, so we now require you to convert trend files if you wish to use the new functionality. Our support cases tell us that this is a good thing.
At this time the new application only has two switches, one to set the source files, and one to write to a single destination file. You can use an unlimited number of input files to write to the new output file and the application will be smart enought to ignore any duplicates. This is cool if you have a series of smaller trend files that you wish to covert to our new format since it is so much better than before. The input files may be the old or the new file format.
Every input file is prefixed with the //it flag and one flag is provided for the //ot output flag as seen below. Returns have been added for clarity.
ncover.upconvert //it c:\mypath\input1.trends //it c:\mypath\input2.trends //ot c:\mypath\output.trends
The command will take the input1.trends and input2.trends file and append the data to a new file called output.trends in the new binary file format.
On a final note. You do not need to convert your files if the xml format works for you. The changes are invisible to the user. NCover and NCover Explorer will continue to read the old and the new format without any intervention from a human.
We would love to hear from you about your trends experience. You can do so by commenting on this blog entry or by emailing us at email@example.com.
Download the new version here.