Is there a better way to watch for new entries in a table besides selecting from it every n ticks of time or something like that?
I have a table that an external program updates very often, and clients can watch for this new data as it arrive, how can I make that without having to set a fixed period of repeatable select statements?
Here's what I do: I've got some triggers set up for the table (insert, delete, update) and those triggers increment a counter in another table. My DB access code keeps a local counter and compares it to the returned value, ultimately sending a bool back to the caller, answering the question IsDataCurrent().
Our programs that use this DB access code either poll or check it on-request and then make the appropriate calls to keep themselves up to date.
I'm sure there are other ways to solve this. It worked for me pretty well, though.
This is just a small improvement to your method. Write a trigger on the table(s) you are watching to update a Last_Changed table.
In MySQL there's no best way than to poll (you create a specific table to simplify the polling though), in other databases you can have triggers that have impact outside the database. In MySQL triggers can only do stuff inside the database itself (for instance, populating the helper table).
Another similar approach would be to add
add column Last_Modified TIMESTAMP ON UPDATE CURRENT_TIMESTAMP DEFAULT CURRENT_TIMESTAMP
to each table and preface your select queries to compare the last request date/time with the max(Last_Modified).
Databases are typically pull sources and not push so you'll still need to programmatically probe for changes no matter what.