Tip 1: Measure execution time of your code
Measuring is knowing. Before you start changing code, make sure you have a set of data you can keep reusing for your tests. Measure the performance of your code on that data after each change in code so you know the impact of your changes.
One way to do this is by using the
WinApiServer::getTickCount() if your code runs on server method.
Tip 2: limit the number of loops
A LOT of time goes into loops. If you have a performance problem, start looking for loops. Code can run really fast, but it can get slow when it is executed too many time, eg, in a loop.
Tip 3: avoid
When there is a
if in a
while select, see if you can rewrite it a a where statement in your select. Don’t be affraid use a join either. Consider the following example:
Tip 4: avoid double use of table methods
Using table methods a lot can get really slow if you do it wrong. Consider the following example:
This example code looks nice, but there’s a problem.
The salesLine.inventDim() method contains the following:
This means that the
invendDim record is read three times from the database. It is better to declare the
inventDim record locally and only retrieve it once:
Tip 5: Don’t put too much code on tables
Code on tables is usually fast, but things can get slow if you use it to much. Say you have a table with an
InventDimId field. If you have 5 methods that need the
InventDim record, because you don’t have a classDeclaration method on your table, you need to call this function 5 times, once in every method:
When you put these methods on a class, you could optimise it by fetching the record only once and storing it in the
classDeclaration, or better, passing it as a parameter to your methods. An other example is fetching parameters from parameter tables, eg
InventParameters::find(). On a table, you have to fetch it each time you call a method. In a class, you would probably optimize your code to only fetch the parameter record once.
Tip 6: Use the fastest code
For some tasks, there is special code that is faster than the code you would normally write. For example:
The same applies to update_recordset for updating records. Also, when adding values to the end of a container
is faster than
Tip 7: Every optimization counts
Remember that every optimization you do to you code counts, even if it’s a little one. Small performance tweaks can have a huge effect once you process large quantities of data. So don’t be lazy, and optimize.
Share from http://www.artofcreation.be/