fetch_row or one by one?

Anon

Greetings

I am writing a script where I need to access several fields of a database in the same row, I wrote it in a way so I would get the data only if need it, if I need two fields then I get them.. then the other two and if I need another field I get the result of it.

However as the script gets bigger the debbuging gets more difficult. It would be less painful if I fetch the complete row right away, no matter if I need some fields or not

Should I really fetch the complete row or should I continue reading field by field? what is faster or what uses more memory?

Anon

Personaly I do it by starting out the script getting all the information from the table, then when I've completed the script I then go back and find what I really needed in the end and then modify the queries to only obtain that information. But I guess it's all in which methood you use with your data.

First pass should always be to get the script to work, second pass is bug removal, third is optimization.

Anon

Good point.
Get the script to work first, then worry about performance.

I don't think you will notice the speed of just selecting what you need instead of fetching the hole row unless you are doing huge amounts of queries or you work with huge rows.

Why not do a test , use the microtime() to get the system time before and after you run the query (including any stuff you do to decide what to select that you would not have to do if you fetched the whole row) and see how long it takes...

Anon

Yeh I've done that, and yep it all depends on how big the return data is, and where your server is located. If it is local then the speed won't be effected too much, but if it is a remote database server then the speed will be greatly increased. Another factor is the server load, how many people are accessing it at any given moment, we ran into this problem when I first programmed up the news page for my site (oglchallenge.otri.net), we'd have something like 10,000 people visit the site in a day and the server just went nutz and was slow as hell all day, after that I went and selected only the data I needed and we havn't run into that problem since.

So it's all up to those 3 little factors, data size, server location and load.