Check for the error returns on everything. Spend time in a profiler, and optimize only the parts that matter. (boring code tends to run on limited systems so optimization is important!)
Trying to handle every error can also be a distraction though. I would suggest to only try to display errors that are likely to occur. It’s okay for a few unlikely errors to be not surfaced other than with a cryptic “there was an error.” This can shorten development time by a lot depending on your architecture.
No, boring software checks for all errors. It might only log the unlikely ones, but it has a log for them all so that if any happen that shouldn't we can figure out why.
Comments
Check for the error returns on everything. Spend time in a profiler, and optimize only the parts that matter. (boring code tends to run on limited systems so optimization is important!)
Trying to handle every error can also be a distraction though. I would suggest to only try to display errors that are likely to occur. It’s okay for a few unlikely errors to be not surfaced other than with a cryptic “there was an error.” This can shorten development time by a lot depending on your architecture.
No, boring software checks for all errors. It might only log the unlikely ones, but it has a log for them all so that if any happen that shouldn't we can figure out why.