notesthoughts

To say it is faster, I need a “before”

The optimization code remained, but records comparing before and after were scarce. Next time, I want to record reproduction conditions and a baseline before changing the code.

My work history shows efforts to reduce rendering in map markers and tables and adjust how frequently chat events trigger updates. I can explain what I changed. But when asked how much better it became, I did not have enough material left to answer.

Having optimization code is different from having evidence of improved performance. Even if I remember why I applied memoization or debounce, it is hard to compare under the same conditions later without a record of the previous state.

Record the same situation first

Next time, I want to begin by describing the specific action that feels slow. Is it changing the selection on a map with a certain number of markers, scrolling a table of a particular size, or sending several requests for one new message? A reproducible situation lets me try the same action after the fix.

Then I can choose what to measure. For a slow click response, I might inspect elapsed time or rendering scope. For refetches piling up with each event, I might count requests. I do not need every metric at once, but I need one baseline that explains the problem.

Even if exact numbers are difficult to collect, I can preserve before-and-after captures or reproduction conditions. That leaves something to revisit instead of relying on memory to say it became faster.

If delaying updates reduces requests, I also need to check whether new messages appear when they are needed. Lower cost alone does not mean the user’s task still works well.

I do not want performance records to become elaborate reports. It is enough to record which situation was slow, what changed, and what differed under the same conditions. For the next optimization, I want to start with that short record before editing the code.